Gender differences in introductory university physics performance: The influence of high school physics preparation and affective factors

Gender Differences inIntroductory University PhysicsPerformance: The Influence ofHigh School Physics Preparationand Affective Factors

ZAHRA HAZARIHarvard-Smithsonian Center for Astrophysics, Cambridge, MA 02138, USA

ROBERT H. TAICurry School of Education, University of Virginia, Charlottesville, VA 22904, USA

PHILIP M. SADLERHarvard-Smithsonian Center for Astrophysics, Cambridge, MA 02138, USA

Received 24 July 2006; revised 1 April 2007; accepted 6 April 2007

DOI 10.1002/sce.20223Published online 29 May 2007 in Wiley InterScience (www.interscience.wiley.com).

ABSTRACT: The attrition of females studying physics after high school is a growing con-cern to the science education community. Most undergraduate science programs requireintroductory physics coursework. Thus, success in introductory physics is usually necessaryfor students to progress to higher levels of science study. Success also influences attitudes; iffemales are well prepared, feel confident, and do well in introductory physics, they may beinclined to study physics further. This quantitative study using a hierarchical linear modelfocused on determining factors from high school physics preparation (content, pedagogy,and assessment) and the affective domain that predicted female and male performance inintroductory university physics. The data analyzed came from 1973 introductory universityphysics surveys that included variables used as controls for student demographic and aca-demic background characteristics. The results highlight high school physics and affectiveexperiences that differentially predicted female and male performance. These experiencesinclude learning requirements, long-written problems, cumulative tests/quizzes, father’sencouragement, and family’s belief that science leads to a better career. There were alsofactors that had a similar effect on female and male performance among which mathemat-ics preparation was the overall strongest predictor of university physics performance. C©2007 Wiley Periodicals, Inc. Sci Ed 91:847 – 876, 2007

Correspondence to: Zahra Hazari; e-mail: [email protected]

C© 2007 Wiley Periodicals, Inc.

848 HAZARI ET AL.

INTRODUCTION

The way science is practiced is largely a product of the way science has developed andsince science was primarily developed by males, it is strongly enmeshed with practicesthat favor male success (Gilbert & Calvert, 2003; Hazari & Potvin, 2005; Lederman, 2003;Wallsgrove, 1980). Lederman (2003) writes

Science is hegemonic and androcentric, two characteristics that proceed from the fact thatpractitioners of science as we know it have traditionally been white, male, and Western. Itis they who define the rules, methods, instrumentation, descriptions of results, and criteriafor knowledge production. It is they who define what counts as science, both theoreticallyand in practice. It is they who are the gatekeepers for access to, and definers of, a life inscience. (p. 604)

Among the sciences, physics is the most extreme in its male dominance. In the 2001–2002academic year, women earned 62% of U.S. bachelor’s degrees in biology, 49% in chemistry,47% in mathematics, but a mere 22% in physics (Snyder, Tan, & Hoffman, 2004).

The historical dominance of males in physics translates into educational practices bydefining what physics content (e.g., topics such as mechanics and electromagnetism) andmethods (e.g., type and format of problems, labs, contexts) are considered suitable forstudying in high school physics. In other words, this content and methodology is derivedfrom the practices of the field itself which have long been biased in one direction. Thus,some of these content areas and methodologies may not be conducive to gender equity interms of performance and interest. Thus, in order to effect change and further diversificationof the field, the educational factors that positively and negatively influence male and femalephysics performance and interest must be identified. This study focuses on examining thefactors that influence performance. This focus is important since performance is the mainmeasure of aptitude used to evaluate educational attainment and performance levels act asa gatekeeper to opportunity. Success also influences attitudes; if females are well prepared,feel confident, and do well in introductory physics, they may be inclined to study physicsfurther thereby entering the field in larger numbers and beginning the diversification andpossible reinvention of the field.

Why Diversify?

The first question to ask is: Why would we want performance, interest level, and par-ticipation in physics to be gender balanced? One rationale is that heterogeneity rather thanhomogeneity will lead to progress in the field by introducing new perspectives. Kenway andGough (1998) observe that the intellectual potential of females is an untapped source forfurthering scientific knowledge. Science will only suffer if there are factors that impede theparticipation of any particular group. In addition, by including a larger cross-section of thepopulation in physics studies, it is reasonable to expect that public interest in physics willincrease as well, especially in the case of females since they make up half the population.Lastly, the pursuit of science is a highly profitable enterprise in our society in terms ofmoney (e.g., jobs), status, and influence (e.g., decision-making capability). Every memberof society should have the capability to economically, socially, and politically empowerthemselves regardless of gender, race, or class. Thus, a clear path should be available towomen who want to study physics, which is currently not the case given the social andeducational barriers they must overcome in order to reach even the lower echelons of thescientific hierarchy. Urry (2003) writes “in physics departments around the country, womenare feeling ill at ease, out of place, not at home” (p. 12).

Science Education DOI 10.1002/sce

GENDER DIFFERENCES IN PHYSICS PERFORMANCE 849

Given these rationales, the second question is: What can we do as educators to achievegender balance in physics performance, interest level, and participation? “Working out thepractical implications of a new approach to content, pedagogy, and assessment methodstakes time and experimentation” (Seymour, 2001, p. 86). This study focuses on answering apart of the question by identifying factors from high school that influence male and femalephysics performance in university. It is important to understand the efficacy of physics-teaching practices in preparing students for a possible future in science in order to givethem the full opportunity to pursue and succeed in science programs.

The Influence of Curriculum and Instruction

Gendering of science is symbolic in that the predominantly remote and abstract presentationof physical science is associated with the masculine, whereas more “humanised” biologicalsciences are often constructed as feminine or gender-ambiguous. This gender symbolisminforms curriculum discourses and practices which in turn reproduce and legitimize genderdivisions. (Hughes, 2001, p. 276)

Among the most important considerations when formulating curricula for a disciplineis deciding what content areas to cover and how to teach those content areas. Althoughteachers have limited choice as to the topic areas they must cover, they have considerablefreedom to decide how to cover these topics. Within the paradigm of traditional physicspedagogy, there are several examples of pedagogical techniques that have been shown tobe unsupportive or even detrimental to female attitude and performance. Among the mostoften cited are the following:

• Teachers who allow male students to dominate classroom interactions, discussions,and activities (Stadler, Duit, & Benke, 2000). “In typical classroom activities,boys often dominate and girls receive less experience” (Chambers & Andre, 1997,p. 118). This is of particular concern since females enter into the classroom with lessprior knowledge and fewer prior experiences with physics (Bell, 2001; Chambers &Andre, 1997; Jones, Howe, & Rua, 2000).

• Unsupportive cooperative learning situations. Laws, Rosborough, and Poodry (1999)found that women in the physics laboratory “complained of domineering partners,clashes in temperament, being subjected to ridicule, fears that their partners didn’trespect them, and feelings that their partners understood far more than they” (p. S35).Another study that examined physics problem solving in cooperative three-persongroups found that “groups comprised of two males and one female tended to bedominated by the male students. . . even when the female member was articulateand the highest ability student in the group” (Heller & Hollabaugh, 1992, p. 641).Despite these negative examples of group work, the role of interaction and cooperativelearning can be a strongly positive one if the group work is designed well andthe interaction is moderated in an appropriate gender-fair way. “Cooperation ingroups is not a strategy per se for a girl friendly physics instruction but only if thegroup work helps communicating and eliciting preconceptions” (Labudde, Herzog,Neuenschwander, Violi, & Gerber, 2000, p. 153). Baker and Leary (2003) found thatthe girls in their study “expressed strong feelings for more interaction with their peersin their repeated requests for group work, partners, and more discussion” (p. 182).

• More topics of interest to males. Haussler and Hoffmann (2002) assert that the physicscontent interesting to girls is almost always interesting to boys but the reverse is not


850 HAZARI ET AL.

necessarily true and that the content interesting to girls is “by far underrepresented inthe curriculum” (p. 872). Similarly, Stadler et al. (2000) report that physics “contextsthat are meaningful for girls are usually also meaningful for boys, though the reversedoes not hold” (p. 417).

• Lack of student engagement. The traditional high school physics class tends tofollow a model of isolationist pedagogy with an excessive amount of reliance ontextbooks and rote problem solving, even though these types of isolated learningmodels have been found to be detrimental to the success of students later on incollege and university physics (Sadler & Tai, 2001). Even when interactive methodsare employed, they often follow “cookbook techniques” and lack the ability to activelyengage students. From their study of elementary education students in a gender-reformed physical science course, Roychoudhury, Tippins, and Nichols (1995) assertthat “ownership is imperative for bringing in those who have been marginalized andthose who have felt that their ability to learn science was deprecated” (p. 916).In introductory university physics, Marshall and Dorward (2000) found that femalestudents and female preservice teachers who participated in inquiry activities showedincreased understanding of physics concepts compared with their peers who hadreceived no inquiry training. However, high school females have been found toresist some types of inquiry (e.g., active physics) and support the traditional physicsparadigm because it allows them to maintain their “good student identities” (Carlone,2004, p. 410).

Research also shows that compared to females in coeducational physics classes, femalesin single-sex physics classes can have higher interest levels, self-concept, confidence,achievement, and persistence (Gillibrand, Robinson, Brawn, & Osborn, 1999; Haussler &Hoffmann, 2002). Parker and Rennie (2002) found that “single-sex classes provide envi-ronments in which teachers can implement gender-inclusive science instructional strategiesmore readily,” separately address boys’ shortcomings in written and oral communicationand girls’ experiential shortcomings, and eliminate the harassment that inhibits girls’ learn-ing (p. 894). However, given that teachers are limited in their ability to separate their classesby sex, the single-sex research at least provides a guide as to what female students mightneed changed in their physics class in order to succeed; that is, alleviating the dominatingrole of male students, gender-inclusive instructional strategies such as “co-operative groupproblem-solving, projects focused on everyday issues, and collaborative practical work”(Parker & Rennie, 2002, p. 894), teacher behaviors to promote positive self-concept andconfidence, and topics of interest to females (Gillibrand et al., 1999; Haussler & Hoffmann,2002).

Another dimension to consider in addition to content and pedagogy is assessment.Biases in physics content and pedagogy can translate into assessment biases just as as-sessment biases can dictate partiality in content and pedagogy when teachers are preparingstudents for a test. Thus, the role of assessment is an important consideration when exam-ining the causes of gender differences in science achievement. Dimitrov (1999) found thatonly among high-ability fifth graders did boys outperform girls and only on open-endedphysical science items (with fixed grading schemes) from a standardized test. No signif-icant differences were found for low- and medium-ability students. At the college level,Hazel, Logan, and Gallagher (1997) found that males outperformed females on traditionalforms of physics questions, especially multiple-choice questions. However, more flexiblequestions led to greater gender equity as did questions that were “decontextualized” inthe sense of having neither a male- nor a female-biased context. Supporting the impor-tance of context, McCullough (2004) found that gender performance differences for some



questions on the Force Concept Inventory test were reduced when stereotypic male con-texts (rockets, cannonballs, hockey, and male figures) were replaced with stereotypic femalecontexts (shopping, cooking, jewelry, stuffed animals, and female figures). Zohar and Sela(2003) divided students’ final physics scores into teacher-given scores and standardizedtest scores and found that boys’ test scores were higher than girls’ test scores, whereasgirls’ teacher-given scores were higher. Bell (2001) found that gender differences favoringmales for physics questions and females for human biology questions exist, but only forquestions that involve the retrieval from memory of declarative knowledge and not forthe deployment of procedural knowledge (knowledge of experimental design and statisticsused in practical work). He also mentions the likelihood of attitude and prior experiencehaving an effect (Bell, 2001). In their study of university students, Chambers and Andre(1997) found that gender had a significant effect on performance for direct current conceptsbut that this effect was eliminated when interest level, experience, and prior knowledgewere included in the analysis. Clearly, there are many ways in which assessments cancontribute to gender bias. The goal of this study is to identify the parts of current highschool physics content, pedagogy, and assessment that might level gender performance inintroductory university physics courses so that females are allowed equal opportunities forpersistence in science. The challenge is then to reformulate traditional curriculum in waysthat eliminate the significant disadvantages females face.

The Influence of Affective Domains

“Despite the widespread belief that emotions are a central part of learning and teaching,contemporary work in science education exploring affect is scant” (Alsop & Watts, 2003,p. 1044). Affective domains can have a powerful influence on both attitudes toward sci-ence and performance in science. The stereotypic image of the science student becomingenthralled with science content without affective support is hardly adequate. According toCleaves (2005), the choice of science involves more dynamic considerations including notonly interest and enjoyment but also “knowledge about science occupations and sciencework,” “relationship with significant adults,” “perceptions of school science,” and especially“confidence in their own ability to do science” (pp. 483–484).

Arguably, the affective domains that students are most influenced by come from theireveryday life and science class plays only a small role in their life experience. The most con-tinual and pervasive influence is from students’ home environment. Entering and becomingpart of a science class is like a “cultural border crossing”—a crossing from the subcultureof their family, peers, and world outside class, to one of school science (Aikenhead, 1996;Brickhouse, Lowery, & Schultz, 2000; Costa 1995). This border crossing becomes an easiertransition for students whose family subculture is consistent with science culture or at leastencouraging and positive toward the study of science.

Affective factors may be especially important for females. Bouchard, St-Amant, andDeslandes (1998) in their study of high school students in Quebec found that parentalinvolvement factors predicted school grades similarly for males and females, with affectivesupport being the strongest predictor. Father’s education was also a significant predictorof school grades for males and females. However, parental affective support and father’seducation were stronger predictors for females than for males. Baker and Leary (2003)found that “the few instances in which the girls chose a physical science career wereall based on having experienced that science with a loved one” (p. S189). In addition, theInternational Study of Women in Physics reported that the factor that women physicists citedmost frequently as contributing to their success was the support of their parents, husbands,advisors, professors, teachers, and sometimes colleagues (Ivie, Cuzjko, & Stowe, 2001).


852 HAZARI ET AL.

For females in physics, affective support seems to have a strong influencing role whether itis from a parent, a teacher, or other close person.

This study addresses the affective factors by examining their ability to predict universityphysics performance. These affective factors include encouragement from different people(father, mother, science teacher, other teacher, etc.), the level of family interest in science,having parents with science jobs, and even the family’s attitude toward science. In addition,this study discerns the relative effects on male and female students’ university physicsperformance of affective variables as well as demographic, academic ability, and high schoolphysics curriculum variables. In this respect, it adds a unique and rigorous perspective tothe literature.

METHODOLOGY

The Study

The research question guiding this study is: What high school physics curriculum, in-struction, and affective factors predict female and male introductory university physicsperformance after controlling for university course, demographic, and academic back-ground variables? The benefit of analyzing the effect of such a variety of factors (content,pedagogy, assessment, and affect) simultaneously is the ability to discover their relativeeffects, that is, which variables are the best predictors given that many different types ofvariables have been included in the model. The method also allows the development of amore comprehensive model to explain the differences in performance. This makes sense foranother reason: students experience these variables simultaneously and not independently.Thus, the model should reflect this simultaneous involvement.

Although the research question has a broad scope, the parent study from which this studyis derived collected a wide range of data that makes it possible to address this question.The parent study is entitled Factors Influencing College Science Success (FICSS) fundedthrough the Interagency Educational Research Initiative and administered through theNational Science Foundation (NSF-REC 0115649). The FICSS study focuses on identifyingpredictors of success in university biology, chemistry, and physics through a large-scalesurvey of introductory science students across the United States. The methodology is thatof an epidemiological survey where the researchers rely on the natural variation in theexperiences, background, and decisions of the sample science students rather than forcingcontrolled group experiments.

The development of the FICSS survey was guided by literature reviews, a preliminaryresearch study conducted in 1994 (Sadler & Tai, 2001), and interviews with 20 highschool science teachers and 22 introductory university science professors regarding factorsthat influence student success in introductory university science. Once the survey wasdeveloped, validity was established through student focus group interviews and consultationwith teachers and professors. These interviews and consultations advised the rewriting ofitems on the survey. Reliability was established through a test–retest study with 113participants, which found that 90.7% of the students responded with at least an adjacentchoice and 60.0% responded with exactly the same response on both administrations. Thefinal survey consisted of 66 items that questioned students regarding the content, pedagogy,and assessment methods of their last high school physics course, the levels to which theytook mathematics and sciences in high school, how they performed in those classes, andtheir demographic information. Included in the survey was a section in which the students’professor reported their grade in the introductory university physics course.



A comprehensive list of 4-year universities (N = 2932) from the U.S. Department ofEducation was obtained in the summer of 2002 while the survey was being developed. Arandom list of 150 schools was generated from the comprehensive list. However, since weonly wanted a random list of schools that offered introductory physics, schools that did notoffer an introductory physics course for science majors were eliminated. There were 56schools on the resulting list of schools that offered introductory physics. Professors teachingintroductory physics within these institutions in the spring and fall of 2003 were identifiedand contacted by members of the FICSS team. Once recruited, the professors administeredthe FICSS physics survey to their introductory physics students and retained the surveysuntil the end of the term. At the end of the term, they completed the final portion of thesurvey by filling in the percent and letter grade received by the student in their course andreturned the surveys to the FICSS team.

The Sample

A concerted effort was made to recruit the participation of at least one professor teachingone section from each of the randomly selected schools. In some schools, we were moresuccessful in recruiting multiple sections and professors. Although we recruited at least onephysics professor from 91% of the schools on our random list (51 of 56), only 80% of therecruited schools (41 of 51) followed through with the administration of the survey, finalgrade entry, and return of the surveys to our group. Of the courses that returned completedsurveys, 88% were first semester physics courses (36 of 41). Thus, the final number ofschools that returned data from first semester physics courses was 36.

The data consist of 3694 surveys, 837 from the spring and 2857 from the fall. Of these,only 2858 surveys were from the first semester courses. As expected, there were missingdata within this data set. Continuous independent variables (a continuous variable takesa value on a scale used to measure it, e.g., grade on a scale of 0 to 100) in this data setwith less than 10% missing were imputed using expectation maximization (EM) in SPSS13.0, which produces maximum likelihood estimates for the missing values that have thegreatest chance of reproducing the observed data. In addition to producing imputed valuesfor missing data, the EM algorithm produces Little’s MCAR test, a form of χ2 statisticthat tests for data being MCAR or “missing completely at random” (Little & Rubin, 2002;Scheffer, 2002; Tai, Sadler, & Loehr, 2005). In the case of MCAR data, the missing valuesdo not appear to have a discernible pattern within the variables being used in an analysisand may be listwise deleted. MCAR is considered a stringent standard for missing data(Allison, 2002; Little & Rubin, 2002; Scheffer, 2002). Little’s MCAR test alerted us to thefact that a variable was not missing completely at random. There are two other situationsfor missing data, missing at random (MAR) and not missing at random (NMAR). Variablesthat are MAR have patterns for their missing values across one or more predictors (e.g.,missing more among minorities than Caucasians, but random within each group) but arerandomly missing across the outcome variable. The patterns in the missing values acrossthe predictors can be used to impute values for the missing data. In our case, MAR variableswith less than 10% missing were imputed. For NMAR, patterns in the missing values occuralso across the outcome and clear biasing would result from any analysis. The purpose forimputation in this case was to avoid the bias due to listwise deletion of variables that wereMAR and to avoid losing large amounts of data (losing statistical power).

Among the students in the sample were those who had taken high school physics aswell as those who had no high school physics experience. However, since the high schoolfactors to be investigated in this study included high school physics content, pedagogy, andassessment, the sample was further narrowed to only those students who had taken high


854 HAZARI ET AL.

school physics. Thus, the final sample consisted of 2085 students, 227 from the spring and1858 from the fall. The remaining missing data (99 missing for gender and 13 missing forthe outcome variable) was listwise removed from the analyses. The final complete data setconsisted of 1973 students within 54 courses from 35 colleges and universities in the UnitedStates. (One university was removed because the class size was 6 students, 5 of whom hadnot taken high school physics, leaving only 1 student in the sample.)

The courses in the final sample were predominantly (83%) fall semester courses for tworeasons: first, the overall fall semester sample was larger than the spring sample and second,first semester physics courses are usually taken in the fall rather than the spring althoughsome schools provide the option of taking first semester physics in either term. The courseswere 59% algebra-based physics and 41% calculus-based physics. Algebra-based physicscourses are often geared toward nonphysics science majors and preprofessional majors(e.g., premedicine, predentistry, prepharmacy, and preveterinary). Calculus-based physicscourses are geared toward physical science majors (physics and chemistry) and engineeringmajors. Both types of courses were surveyed since both types of courses include sciencemajors who need to do well in introductory university physics in order to graduate fromtheir science programs and continue to further science studies. The school location anddistribution of the sample within the schools (estimated course enrollment, number ofsurvey responses, and final sample number) are listed in Table 1.

The estimated course enrollment figures were determined through multiple methods: theprofessor reported the number of students in his/her course, the school’s course schedulewas consulted as to the number of seats taken up in each section of the course, or in caseswhere the former two pieces of information were absent, the enrollment was estimated basedon available enrollment data from courses in other schools of similar size. The response rateaccording to the estimated enrollment was 76%. Of the student responses, 69% were fromstudents who had taken high school physics. These percentages are on par with a similarearlier study (Tai & Sadler, 2001). There were 12 private schools and 23 public schools inthe sample. This is not problematic since a much higher percentage of students in the UnitedStates attend public schools than private schools. Figure 1 shows the regional distribution ofschools by school governance and size. There was a fair spread of colleges and universitieswithin the regions of the United States, although the sample from the northeast was smallerthan the sample from other parts of the nation. The number of schools in each size rangewas nicely distributed so that there was an almost equal representation of each size school.The average SAT admission score across all the schools was 1067 of 1600. All the largeschools had average SAT admission scores of 1000 and above. Some small and mediumschools had SAT scores less than 1000, with 830 being the lowest average score.

Controls, Predictors, and Outcome

The demographic control variables included highest parental education (control forsocioeconomic status), race (White, Native, Asian, Black, multiracial), ethnicity (Hispanic),and high school governance (public, private). The academic control variables included SATmathematics scores, last high school English, mathematics, and science grade, high schoolcalculus enrollment (regular and advanced placement (AP)), year in university (freshman,sophomore, junior, senior, graduate student, other), and whether the student dropped, failed,or had difficulty with course previously. Finally, the focus predictor variables in this studyincluded gender, high school physics curriculum predictors, affective predictors, and theinteractions of the latter two with gender. More specifically, among the curriculum predictorswere variables such as the time spent on certain content (e.g., mechanics, optics), the levelof understanding of the material required, number of projects, number of labs, number



TABLE 1School Location and Sample Distribution by Class

Estimated Surveys SampleCollege/University State Enrollment Collected Numbera

1 Alaska 30 20 530 18 850 31 2351 46 37

2 Arizona 50 23 21150 113 81173 122 88

3 California 35 23 1373 58 33

4 California 55 23 1254 47 2780 67 4170 54 45

5 Colorado 76 53 366 Georgia 55 55 257 Georgia 85 59 468 Iowa 28 27 189 Idaho 140 105 3710 Illinois 17 16 1111 Illinois 29 24 2112 Illinois 85 78 6413 Indiana 70 52 3114 Indiana 45 37 3415 Kansas 350 317 23116 Kentucky 34 30 18

35 26 2017 Kentucky 16 15 1018 Kentucky 30 27 18

280 191 13719 Louisiana 80 29 21

91 82 5620 Louisiana 30 18 18

68 55 5390 62 55

115 108 9521 Maryland 65 37 34

140 116 5922 Michigan 40 30 2723 North Carolina 22 21 1324 Nebraska 17 17 1025 New Jersey 14 8 726 New York 60 36 2727 Oregon 30 27 24

45 39 3628 Pennsylvania 30 25 2029 South Carolina 20 13 530 Tennessee 18 15 12

Continued


856 HAZARI ET AL.

TABLE 1School Location and Sample Distribution (Continued)

Estimated Surveys SampleCollege/University State Enrollment Collected Numbera

31 Texas 50 23 1332 Texas 8 8 5

50 47 4133 Utah 155 135 6734 Washington 89 61 32

120 67 4335 West Virginia 30 22 9Overall – 3753 2858 1973

aStudents who took high school physics.

of demonstrations, time spent discussing demonstrations, time spent reading, types ofproblems done, pedagogical techniques (e.g., lecture, group work), and types of assessmentquestions. Among the affective predictors were who encouraged the student to take science,degree of family interest in science, having parents with science jobs, and the family’sattitude toward science. For more information on the FICSS survey items, a sample of thesurvey can be accessed from http://www.cfa.harvard.edu/smg/ficss/research/survey.html.

The outcome variable is performance in introductory university physics, which wasreported by the university professor. It was important to reconcile the different gradingschemes used by the professors participating in our study. Some professors reported onlywhole-letter grades (A, B, C, etc.), others used pluses and minuses (A+, A, A–, B+, etc.),and still others used point scales (0–100). All of these different final grading schemes

Figure 1. Number of colleges and universities by governance and size.



were converted to a 100-point scale, where A+ = 98; A = 95; A– = 91; B+ = 88; B = 85;B– = 81, etc. This approach was based on the approach used in previous studies of introduc-tory university performance (Sadler & Tai, 2001; Tai et al., 2005). The final outcome variableis labeled GRDPRCNT and takes a value between 0 and 100. The overall GRDPRCNTaverage for this sample is 80.8 ± 10.1 translating to a grade of B–. Although the gradereceived in an introductory physics course may not be reflective of conceptual understand-ing or even learning, it is a measure of whether a student will be able to continue studyingscience since the large introductory science courses tend to be gateway courses that sortout those who will continue in science from those who will not. Gainen (1995) reportedthat dropout and failure rates were high in introductory “gate-keeping” university courses.Sadler and Tai (2001) write, “While success in introductory physics opens the door forstudents to opportunities in engineering, medicine, and scientific research, failure in thesecourses closes those career options and presses students toward nonscience fields, negatingyears of preparation and aspiration” (p. 112). Thus, performance in these courses is vitalfor maintaining students within science programs.

Modeling

The research question is investigated by fitting a hierarchical linear model (HLM) usingthe MIXED command in SPSS 13.0. As with regression, hierarchical models allow us topredict an outcome based on one or more predictor variables. Since this study has studentsnested within courses, the multilevel model consists of two levels of variation: (1) thestudent level and (2) the course level. (Note that the third level of variation, i.e., schools,was nonsignificant after courses were accounted for.) The purpose of the modeling wasto discover factors from high school physics and the affective domain that influence maleand female students’ introductory university physics performance similarly and differently.In order to extract the factors that might influence males and females differently, genderinteractions were included in the model.

Before building the model, the data set was cleaned and preliminary analyses wereconducted. It was important to understand the level of correlation and intercorrelationbetween the variables so that a variable would not be inappropriately entered or removedfrom the model when its presence was influenced by another variable. These correlationshighlighted potential confounders or effect modifiers that were important to look out forin the modeling as well as potential problems with redundancy. Variables that were evenslightly correlated or intercorrelated were carefully monitored during the model building.Collinearity diagnostics (tolerance and variation inflation factor) were also checked. Inaddition, the independent significance of each individual predictor and the significance ofincluding the predictor (increase in explained variance) as the model was being built werecarefully considered. Predictors and interactions not meeting or close to the minimum levelof significance (p < .05) were excluded from the model.

RESULTS AND DISCUSSION

Descriptives

To better understand the sample, it is important to know how some general student char-acteristics are distributed within the sample. Table 2 provides a summary of the proportionof students in various demographic and academic groups. Gender representation in thesample is not equal. Although Tai et al. (2005) found that females dominated a randomsample of students in introductory university chemistry courses, in this study of introductory


858 HAZARI ET AL.

TABLE 2Frequencies and Percentages of Some Student Characteristics

Demographic/Academic Characteristics Subsample Percentage of Sample (N = 1973)

GenderFemale 780 39.5Male 1193 60.5

Race/ethnicityWhite 1408 71.4Native 32 1.6Asian 211 10.7Black 123 6.2Multi 45 2.3Hispanic 123 6.2

Private high school attendance 405 20.5Previously dropped,

failed, difficulty with course 100 5.1Year in university

Freshman 293 14.9Sophomore 758 38.4Junior 592 30.0Senior 274 13.9Graduate student 22 1.1Other 30 1.5

Calculus enrollmentCalculus 396 20.1AP calculus AB 500 25.3AP calculus BC 162 8.2

physics, female representation lagged male representation by 20%. White students madeup most of the sample (71.4%) with Blacks and Hispanics each representing 6.2%. Mostof the students in the sample had attended public high schools (80%) and reported neverhaving taken introductory physics in university before (95%). Again, unlike Tai et al.’s(2005) chemistry sample, which was dominated by freshmen, this sample was dominatedby sophomores. Around half of the students in the sample reported taking some form ofhigh school calculus.

TABLE 3Means and Standard Deviations of Select Continuous Variables AcrossMales, Females, and the Entire Sample

Female Male TotalSignificant

Mean SD Mean SD Differences Mean SD

Highest parental education 2.8 1.1 2.9 1.1 ns 2.9 1.1Last high school English grade 4.7 0.5 4.4 0.7 *** 4.6 0.6Last high school mathematics grade 4.6 0.7 4.4 0.8 *** 4.4 0.8SAT mathematics score 602.1 91.1 622.4 92.9 *** 614.4 92.7Last high school science grade 4.5 0.7 4.4 0.7 * 4.4 0.7GRDPRCNT 82.3 10.9 81.4 12.0 ns 81.7 11.6

∗, p < .05; ∗∗, p < .01; ∗ ∗ ∗, p < .001.



Table 3 provides a summary of means and standard deviations for select academiccharacteristics. The academic characteristics also include a comparison across males andfemales and the significance level of the gender difference. Although females significantlyoutperformed males on their last English, mathematics, and science high school grades,males outperformed females on the mathematics section of the SATs. This is not surprisingsince females consistently earn lower scores on the SATs than their male counterparts(Leonard & Jiang, 1999; Wainer & Steinberg, 1992). It is interesting to note that althoughfemales are entering university in greater numbers with grades competitive to or even betterthan males (Sommers, 2000), they are still less likely to enroll in physics courses thanare males. For those who do enroll in introductory university physics, their grades are notsignificantly different from the males in the course despite the fact they enter with betterhigh school grades.

Models

The two models fitted include a model with only the academic background variables andone with the academic background, high school curriculum, and affective domain variables.These models are summarized in Table 4. Note that very weak effects (those closest top = .05) were not included in the models to avoid possible Type I errors. The variancein GRDPRCNT is 133.68 (standard deviation squared), and 112.75 of which lies at thestudent level (84%). Thus, the variation in GRDPRCNT due to students being in differentuniversity courses constitutes the remaining 16%. The course-level variation representsa significant (p < .001) portion of the variance in GRDPRCNT. This indicates that theuniversity courses taken by students varied significantly in their grading practices and/orlevel of student. However, this variation was separated from the student-level variation, sothat differences between students’ high school experiences could be determined withoutbeing biased by university course variation.

There is no significant difference in the average university physics grade received bymales and females (see Table 3). Thus, gender alone was found to be a nonsignificant pre-dictor of GRDPRCNT. However, once we controlled for academic background in ModelI, gender became significant. This indicates that although there were no significant differ-ences in male and female performance in introductory university physics to begin with,once academic background was controlled, the difference became significant. Since theparameter estimate for “female” is significant and negative in Model I, this tells us thatfemales are doing significantly worse than males relative to their incoming academic levels.In other words, females come into university with higher grades yet they still perform atthe same level in introductory university physics when they should be performing at higherlevels. Once the curriculum and affective variables were added in Model II, gender becamenonsignificant; that is, after adding the content, pedagogy, and assessment predictors thatinfluence male and female performance differently (significant interactions), the main effectof gender in itself became negligible within errors. Another important observation is thatthe academic background predictors, especially the mathematics background predictors,are the most important predictors in both models.

Model I increases the ability to explain the student-level variation in GRDPRCNT by 21%indicating that students’ academic background plays an extremely large role in predictingtheir university physics performance. In going from Model I to Model II, the percentexplained only increases by 7% (5% explained by the curriculum variables and 2% by theaffective variables). Thus, adding the high school physics curriculum and affect predictorsonly increases the predictive value by a third. In other words, what happens in high schoolphysics is much less important than students’ academic backgrounds and especially their


860 HAZARI ET AL.

TABLE 4Hierarchical Linear Models Predicting GRDPRCNT With a Focus on GenderInteractions

Model I Model IIN = 1973Parameter Estimatea B Significance SE B Significance SE

Intercept 47.51 *** 2.50 47.45 *** 2.73Demographic and academicFemale −1.27 ** 0.48 −2.77 ns 1.43Racially Black −5.09 *** 1.08 −4.81 *** 1.06Ethnically Hispanic −2.19 * 0.95 −2.17 * 0.92Graduate student 7.65 *** 2.19 7.10 ** 2.12Last high school English

gradeb1.04 ** 0.38 1.01 ** 0.37

Last high school mathematicsgradeb

2.80 *** 0.33 2.61 *** 0.32

SAT mathematics scorec 0.016 *** 0.003 0.015 *** 0.003Took regular calculus in high

school1.84 ** 0.60 1.63 ** 0.59

Took AP calculus AB in highschool

2.70 *** 0.61 2.67 *** 0.59

Took AP calculus BC in highschool

3.07 ** 0.90 3.11 *** 0.88

Last high school sciencegraded

1.80 *** 0.36 1.69 *** 0.35

High school physicscurriculuma

4 weeks on mechanics 2.11 ** 0.63A semester on mechanics 2.55 ** 0.81Mechanics a recurring topic 2.64 *** 0.69A semester on optics 5.86 ** 1.82History of physics a recurring

topic1.82 ** 0.67

Learning required:memorize/understand

−0.20 ns 0.26

Number of student-designedprojectse

−0.44 ** 0.17

Read/discussed lab a daybefore

−1.53 ** 0.57

Number of microlabsc −0.15 *** 0.04Number of videos/DVDsc 0.05 * 0.02Minutes discussed demos

afterwardse−0.05 * 0.02

Number of long-writtenquestions’ per weekd

0.12 * 0.05

Test questions requiredcalculations

1.97 ** 0.76

Test questions covered earliermaterial

1.34 * 0.60

Continued



TABLE 4Hierarchical Linear Models Predicting GRDPRCNT With a Focus on GenderInteractions (Continued)

Model I Model IIN = 1973Parameter Estimatea B Significance SE B Significance SE

Affective factorsFather encouraged to take

science−1.34 * 0.61

Family belief: science for abetter career

2.61 *** 0.73

Family belief: science iscourses to pass

−1.40 * 0.54

Interaction with femaleLearning required:

memorize/understand0.79 * 0.38

Number of long writtenquestions per week

−0.22 ** 0.08

Test questions covered earliermaterial

−2.38 ** 0.92

Father encouraged to takescience

2.55 ** 0.93

Family belief: science for a −2.93 ** 1.05better career

Variance componentsStudent level (Rij)f 89.42 *** 2.88 81.74 *** 2.58Course level (U0j)g 21.10 *** 4.84 23.42 *** 5.28

Percent explained (studentlevel)

21% 28%

aAll members of a dummy variable set were included in the model but only significantpredictors are reported.

bLess than 2% imputed.cLess than 10% imputed.dLess than 7% imputed.eLess than 5% imputed.f The unexplained student level variance for the unconditional model (no predictors) is

112.75.gThe unexplained course level variance for the unconditional model (no predictors) is

18.18.∗, p < .05; ∗∗, p < .01; ∗ ∗ ∗, p < .001; ns: not significant.

mathematics preparation. However, Model II is still insightful since it addresses what mayor may not be working in high school physics, however small the effects might be. Thedeviance or “–2 log likelihood” is significantly decreased (p < 0.001) by both Models Iand II indicating that they are better fits than the previous model (i.e., Model I is better thanthe model with no predictors and Model II is better than Model I). The final model, ModelII, explains 28% of the student-level variance in GRDPRCNT.

Positive Predictors

Most of the positive predictors for both males and females are background academicpredictors in the form of mathematics SAT scores, calculus enrollment, and high school


862 HAZARI ET AL.

grades in English, mathematics, and science. Among these academic predictors, the im-portance of mathematics preparation and ability is well documented as being a majorcontributor to success in university. Shumba and Glass (1994) found from a surveyof college faculty appointed to head/coordinate freshman chemistry that high schoolgraduates were insufficiently prepared for college in terms of mathematical skills. Onefaculty member in the study commented: “While basic information in the sciences is im-portant, math through spherical trig. and elementary calculus, and application of those skillsthrough word problems is more important” (Shumba & Glass, 1994, p. 389). A decade later,the Association of American Universities in their 2003 report, Understanding UniversitySuccess, stressed that for university-level success in the natural sciences, mathematics skillsare crucial. These skills include knowing and understanding “basic mathematical conven-tions,” “basic trigonometric principles,” “the relationships between geometry and algebra,”and mathematics as a symbolic language. Additionally important is the ability to “recognizeand use basic algebraic forms,” “work algebraically with formulas and symbols,” “prob-lem solve,” and apply concepts of probability, statistics, and measurement (Conley, 2003,pp. 42–44). One faculty member commented: “Basic math skills are, quite possibly, themost important set of skills for students to have mastered coming into a freshman sciencecourse. They need to understand why equations work and what each equation says aboutthe physical world. . . ” (Conley, 2003, p. 40). In addition, previous quantitative studies havefound that high mathematics SAT scores (Long, McLaughlin, & Bloom, 1986), calculusenrollment (Sadler & Tai, 2001), high-mathematics grades in high school (Alters, 1995; Gif-ford & Harpole, 1986; Hart & Cottle, 1993), and high overall grades in high school (Sadler& Tai, 2001) are associated with high performance in introductory university physics.

Although overall high school achievement has been found to influence university physicsperformance, the influence of specific areas other than mathematics and physics is largelyunexplored. It is surprising that even after including SAT mathematics scores, calculusenrollment, and last high school mathematics and science grade in Model II, the last highschool English grade was still a significant predictor of performance. To understand themeaning of this result, consider the example of two students with the same demographicbackground and high school mathematics and science scores who attend the same universitycourse. These two students, however, have different English achievement levels—one gotan “A” in his/her last English class and the other got an “F.” According to Model II, thefirst student would score approximately 4% points (almost half a letter grade) better inintroductory university physics than the second even if both scored “A” in their last highschool mathematics and science courses. This result indicates the importance of readingand writing skills not taught in a mathematics or science class to science performance.In fact, reading and writing analytically have been found to be important strategies forprocessing scientific information (Musheno & Lawson, 1999; Rivard & Straw, 2000).

The next largest group of positive predictors consists of high school physics contentpredictors. The content predictors consisted of six dummy variable sets indicating theamount of time (0 weeks, 2 weeks, 4 weeks, 18 weeks, or recurring topic) spent on thesix content areas of interest: mechanics, electromagnetism, optics and waves, heat andkinetic theory, relativity, and history and people of physics. Since “2 weeks” was themost frequently reported time for all of the topics, it was excluded from the model. Thus,the results for the other time frequencies are compared to the “2 weeks” standard. Forexample, spending “4 weeks” on mechanics is highly significant in Model II, indicatingthat a prototypical student who spent 4 weeks on mechanics in high school does 2.13%points better in university physics than the equivalent student who only spent 2 weeks onmechanics. The positive mechanics predictors include spending 4 weeks, a semester, orrecurring time throughout the year on mechanics instead of just 2 weeks. Also predictive



of higher performance in university physics was spending a semester on optics and waves.These results point to the benefit of covering topics for longer amounts of time. This result isconsistent with the results of Sadler and Tai (2001) and the Third International Mathematicsand Science Study (TIMSS) (Forgione, Nohara, Welch, & Salganik, 2000). In response tothe TIMSS results, the National Science Teachers Association (NSTA) strongly advocated“teaching less content at greater depth (the principle of ‘less is more’)” to enhance studentperformance and learning (Rakow, 1999, p. 32). Following suit, the American Associationfor the Advancement of Science now also advocate “reducing the number of major topicstaught, pruning unnecessary details or subtopics, de-emphasizing technical vocabulary andeliminating repetition” as major goals (Fratt, 2002). However, change to the “less is more”curriculum has been slow and the education system is still fraught with physics classes thattry to cover the entire textbook to prepare students for university.

The last positive high school content predictor is having history of physics as a recurringtopic. It is encouraging that the way physics teachers incorporated the history and peopleof physics into their high school physics course positively predicted student performancein university. It is likely that the influence was an indirect one where the history topicscontributed to a number of domains such as humanizing science, refining critical thinkingskills, promoting a deeper understanding of topics, addressing misconceptions (Matthews,1994; Rudge & Howe, 2004), and bringing current scientific conceptions into perspective(Monk & Osborne, 1997), which subsequently influenced performance. However, the in-clusion of the history of physics is contentious because though it may positively influenceperformance, it may also present an elitist and singular view of the definition of physics,how it is pursued, and who the practitioners are, leading to the alienation and disinterestof students who do not “fit the mold.” Thus, perhaps a movement toward wider Science,Technology, and Society inclusion that incorporate aspects of the history of physics is thebest direction in which to move.

Among the high school pedagogy predictors, there is only one significant positive pre-dictor for both males and females: the frequency of watching videos or DVDs. In support,Harwood and McMahon (1997) found that

. . . integrated video media curriculum intervention can positively affect student chemistryachievement and attitude across ability levels and across a diverse multicultural population.Furthermore, the data suggest that educational science video media in general, and theWorld of Chemistry video series in particular, are instructional tools that can be usedeffectively to bring the often abstract, distant worlds of science into close focus and withinthe personal meaningful realm of each individual student. (p. 617)

TV, videos, and DVDs are arguably among the most powerful mediums used to influencethe thoughts, ideas, and conceptions of people in our society. As demonstrated by Model II,this medium can also be used in a science class to influence future performance of students.With the advent of excellent science programming for inclusion in science lessons offeredby “The Annenberg Channel,” “The Discovery Channel,” “NOVA,” and countless others,it is not surprising if students responded to video as a medium for science education.

The only significant high school physics assessment predictor that positively influencedmale and female performance was having test questions that involved calculations. Typicaltest questions in university physics are problem-solving questions that involve calculations.Thus, for the high school physics students in this study, the inclusion of test questions withcalculations can only have served to prepare them for the types of questions they would seein university physics.


864 HAZARI ET AL.

Negative Predictors

The greatest quantity of negative predictors for both male and female performance isfound within the “high school physics pedagogy” domain. The positive predictors, onthe other hand, have only one member in the pedagogy domain (number of videos). Thefirst negative predictor is the number of student-designed projects done in high schoolphysics. In the past decade, a strong movement in support of student-designed projectsas a viable pedagogy has arisen within the science education world. The purpose of theseprojects is to actively engage students in the process of learning science by allowing themdifferent levels of control and ownership over the problem, process, and product (studentcentered), thereby embedding the project in a context relevant to the student. Schneider,Krajcik, Marx, and Soloway (2002) found that students in project-based science (PBS)classes on average scored higher than a national sample on long response questions andscientific investigation questions from the National Assessment of Educational Progress(NAEP) science achievement test. However, there was not much difference on the practicalreasoning questions. In university physics, the questions students will typically encounterare problem-solving questions, often multiple choice, that require practical reasoning usingconcepts learned. They do not usually encounter long response questions or questionsconcerning scientific investigations/process. Thus, the advantage for students in Schneideret al.’s (2002) study would not be an advantage in university physics. In addition, mostPBS classrooms require revised curriculum and assessment (Barron et al., 1998). Thus,although student-designed projects might increase student interest and engagement, theyrequire substantial curriculum change for effective implementation—an implementationthat may have the potential to positively influence university physics performance. Theresults of this study do not indicate that student-designed projects are negative predictors ofuniversity performance in all cases. However, the results do indicate that the way they arecurrently being implemented in most classrooms is a negative predictor. A further point ofclarification is that the results say nothing about interest or attitude—such projects mightpositively influence students’ attitudes.

Three of the negative pedagogy predictors relate to labs and demonstrations in highschool physics. They are as follows: read and discussed labs in class a day before doingthe lab, frequency of microcomputer-based labs (MBL), and time spent on discussion aftera demonstration. It is likely that all of these negatively influenced performance becausethey took up considerable class time with little or no gain in understanding or developmentof skills needed in university. In the first case, a teacher might exhaustively go throughthe details of the experimental theory and procedures the day before the lab, leaving littleopen-ended possibilities for student exploration. The purpose of a lab is to enhance stu-dent mastery or conceptual understanding, develop scientific reasoning or practical skills,exemplify the nature and complexity of scientific work, cultivate interest, or develop groupwork skills (Singer, Hilton, & Schweingruber, 2006). All of these goals suffer if the lab isdrawn out in too much detail beforehand because students are not given the opportunity to(i) struggle with the concepts, (ii) figure out how to address the material and use/design theequipment, (iii) feel frustration, (iv) develop curiosity, or (v) deliberate, brainstorm, andmake decisions with others. Many of the tasks that involve deeper conceptual understandingor higher order thinking skills that might benefit students’ future work in science may becurtailed when the lab is too extensively and systematically presented. In addition, valuableclass time is lost. This is not to say that introducing labs is always negative but rather thatintroducing the lab extensively the day before was a negative influence on performance forthis random sample. In other words, the finding is indicative of a problem in the way labsare generally being introduced on the day before they are undertaken. Tai et al. (2005)



corroborate this result with an identical result relating to high school chemistry labpreparation.

The second case is MBL frequency. Microcomputer-based labs involve capturing, analyz-ing, and displaying real-time data in science experiments using tools such as amicrocomputer, probes, graphing software, etc. Kulik (2002) summarizes some of theresearch on MBLs as follows:

Reviewers who examined the early evaluation literature found few studies that showedlearning advantages for MBL instruction. . . Effects of Using Instructional Technology(NSF report) also found no consistent MBL contribution to student learning. . . studentswho learned in MBLs performed no better on tests than did students who learned inconventional laboratories. (pp. 4–5)

If MBLs do not improve student learning, performance, or interest, then their high fre-quency in high school physics is a waste of class time and subsequently a detriment tostudents’ performance. Supporting the above research results, we found that increasingMBL frequency in Model II results in a linear decrease in performance, that is, the wayMBLs are broadly being used by high school physics teachers is a negative predictorof university performance. For example, consider three equivalent prototypical students.The first reports not having any MBLs, the second reports having an MBL five timesin a semester, and the third reports having an MBL 10 times in a semester. Accordingto Model II, the second student will do 0.75 points worse than the first and the thirdwill do 1.5 points worse than the first in university physics. However, this is not to saythat MBLs could not be implemented in such a way that they would benefit future per-formance. For example, Redish, Saul, and Steinberg (1997) found that MBLs benefitedstudents’ understanding/performance in an introductory university physics course for en-gineers when the MBLs were implemented with an active engagement/discovery com-ponent. They further hypothesized through anecdotal evidence that the use of the sameMBL equipment in a traditional lab format would fail without the engagement/discoverycomponent.

In terms of the last case, traditional demonstrations in physics that involve students ob-serving and listening to explanations have been found to have little to no effect on physicslearning and/or performance (Crouch, Fagen, Callan, & Mazur, 2004; Roth, McRobbie,Lucas, & Boutonne, 1997). However, demonstrations can be more effective when studentsmake predictions, record their predictions, discuss with fellow students, and then hear theexplanation (Crouch et al., 2004). Although Crouch et al. found discussion after demon-strations to be effective, this discussion was paired with students making predictions andwriting them down at an earlier time. If discussion is undertaken after a traditional demon-stration without predictions, the effect will likely be reduced. In addition, Crouch et al.’sstudy involves discussion with two or three peers and not a whole-class discussion. In gen-eral, the most important component to the efficacy of demonstration is predemonstrationpredictions and actively engaging students. The debriefing/explanations after demonstra-tions are less important and might cut into other more valuable lesson components. Thus,in Model II, time spent on discussion after a demonstration linearly decreases performancesuch that two students who report spending 15 and 30 minutes, respectively, on discussionafter a demonstration do respectively 0.75 and 1.5 points worse than a student who reportsno time on discussion.

The last predictor that negatively influenced performance in university physics belongsto the affective domain. It is whether students reported that their family viewed science as“a series of courses to pass.” It is not surprising that such a dismissive family attitude toward


866 HAZARI ET AL.

learning would negatively affect student performance. Andre, Whigham, Hendrickson, andChambers (1999) assert,

Importance of a subject area to the child’s future as perceived by the parents may haveprofound implications. Perceived importance may directly affect the amount of encour-agement a parent would provide to the child and the opportunities provided to the childthat may be manifest in the type of activities, toys, and reading materials provided. As achild becomes older, importance is undoubtedly related to the belief of certain subject areasbeing important because they are important for future jobs. (p. 742)

Research supports this claim. Smith and Hausafus (1998, p. 121) found in all of theirregression models that mothers reporting “My child needs to know only a minimum amountof science; therefore, advanced science courses are a waste of time for him/her” was relatedto lower performance of minority eighth-grade students on standardized mathematics andscience tests. Labudde et al. (2000) found that parents’ attitudes toward physics and theirexpectations of their 11th-grade child’s ability in physics were highly related to the child’sperception and expectation of their own physics ability as well as their actual achievement.According to Model II, prototypical students who report that their family believes thatscience is a series of classes to pass do on average 1.4 points worse in university physics.

Gender Interactions

Thus far, the predictors that have been summarized have a similar influence on maleand female performance in university physics. This section details the predictors that havea significant interaction with gender, that is, they predict male and female performancedifferently. In total, there are five of these interactions in Model II (see Table 4). Two arefrom the high school physics pedagogy domain, one from the assessment domain, and twofrom the affect domain. There are no significant interactions from the background or highschool physics content domains.

The first predictor from the pedagogy domain to interact with gender is the high schoolphysics course’s learning requirements, that is, whether the course required “a lot ofmemorization” (on one end of a 5-point scale) or “a full understanding of topics” (on theother end of the scale). However, the effect is a small one (p < .05). High school physicscourses that required a full understanding of topics seemed to benefit female students morethan male students in university physics. On the other hand, high school physics coursesthat required memorization seemed to benefit male students more than female students.Figure 2 illustrates the predicted shift in university physics points by high school physicslearning requirements and gender according to Model II. Moving up the y-axis indicates anincrease in university grade and right along the x-axis indicates an increase in the level ofunderstanding required as opposed to memorization in high school. The gender result is notsurprising after considering the effect of prior experience and knowledge as well as studyand work habits. Several studies support the finding that males have considerably more priorexperience and knowledge of physics coming into a science class (Bell, 2001; Chambers &Andre, 1997; Jones et al., 2000; Roychoudhury et al., 1995). However, females have beenfound to be more studious and diligent toward schoolwork (Barker, 1997; Sommers, 2001;Xu; 2006). In addition, research has found “academic work” (studying, doing homework,memorizing—not achievement) to be perceived as “feminine” by boys, resulting in boysavoiding exhibiting such work behaviors (Epstein, 1998, Jackson, 2002). Clearly, Model II isexhibiting what female and male students are lacking and would subsequently benefit from.In other words, female students lack prior experience and conceptual understanding andthus benefit more from a high school physics course that focuses on those aspects. Males,



Figure 2. Shift in university physics grade percent (GRDPRCNT) predicted by gender and learning requirementsin high school physics.

on the other hand, lack perseverance in studying and committing practical knowledge(definitions, equations, specific techniques) to memory. Thus, they benefit from a coursethat forces them to develop such habits and skills.

Clearly, there is a balance that needs to be struck that accounts for both female and maledeficits coming into high school physics. One way that has been found to be particularlyfruitful in increasing female understanding is exposing them to real-world examples. Mostfemales understand physics better when they can put concepts into a “broader world view”(Stadler et al., 2000). In addition, females tend to enjoy the application of broader worldperspectives, such as applications to people and society, within the study of science (Baker &Leary, 2003; Jones et al., 2000; Lie & Bryhni, 1983). In fact, for our data, the survey itemasking students about the frequency of real-world examples in their high school physicsclass was highly correlated (p = .000) to the learning requirements item. Thus, studentswho reported that their high school physics class was focused more on understanding alsoreported a higher frequency of real-world examples.

The second significant gender interaction within the pedagogy domain is the number ofproblems with long-written explanations that students reported doing every week. Unlikethe last two interactions, this illustrates the opposite trend with respect to gender wherefemales who did long-written problems performed worse than their male counterparts.Figure 3 demonstrates the relationship where moving up the y-axis indicates an increasein university grade and right along the x-axis indicates an increase in the number oflong-written problems students had to answer each week in high school physics. Typicalproblems that students in university physics are required to solve include those requiringcalculations, understanding mathematical relationship between variables, and/or conceptualunderstanding. Questions that require long-written explanations are virtually nonexistentin introductory university physics courses. Thus, excessive use of long-written questions inhigh school might not prepare students for the types of problems they will have to solve in


868 HAZARI ET AL.

Figure 3. Shift in university physics grade percent (GRDPRCNT) predicted by gender and number of long writtenproblems per week in high school physics.

university and subsequently lead to underperformance. Or perhaps the types of long-writtenproblems that are currently being used are not particularly helpful in building knowledgeor skills helpful in university. This explains the result for females. However, for malesthe relationship is strongly positive. Males in the past few decades have fallen far behindfemales in their reading and writing skills (Sommers, 2001). The largest achievement gapbetween males and females is now in writing (Freeman, 2004). Thus, males who wererequired to answer physics questions with long-written explanations may have been gettingpractice in necessary literacy skills they lacked. Supporting this result, Rivard and Straw(2000) in their study of eighth-grade science students, found

. . . that boys who wrote after discussing the problems with peers showed better retention offacts and simple concepts over time than other boys who either worked alone on descriptivetasks or just discussed possible explanations with others without writing. On the other hand,girls who had the opportunity to discuss the problems with peers showed better retentionof simple knowledge over time than other girls who just wrote in response to the problemtasks. (p. 586)

Thus, it is likely that writing tasks are more beneficial to male performance than femaleperformance.

The only assessment variable that significantly interacts with gender to affect perfor-mance is having test questions that included “material covered on previous tests/quizzes.”This dichotomous variable is positively related to male performance and negatively re-lated to female performance. Figure 4 illustrates the predicted shift in male and femalestudents’ university physics points when they report whether they had cumulative highschool physics tests. Since males have greater prior experience in physics (Chambers &Andre, 1997; Jones et al., 2000; Labudde et al., 2000), they come in with a stronger



Figure 4. Shift in university physics grade percent (GRDPRCNT) predicted by gender and tests covering materialfrom earlier tests in high school physics.

knowledge base and their gains in conceptual understanding are probably more substantivethan the females. Williams (2000/2001) found that introductory physics students who hadhigh levels of “communication apprehension” were more prone to rote learning such asmemorization and that females had significantly higher levels of such “communicationapprehension.” Cavallo, Potter, and Rozman (2004) found that females shifted toward morerote-learning strategies by the end of a university physics course, whereas males shifted tomore meaningful strategies. If females in the sample were relying more on memorizationthan understanding, then their learning was superficial at best. Thus, for them, taking cu-mulative tests would have required assigning much more information to memory withoutconceptual understanding. In addition, assigning large amounts of information to memory(cramming) rather than understanding the material conceptually puts the student in dangerof the obvious: forgetting! This would not be a problem for the males who enter physicsclass with some conceptual understanding and then build upon it while avoiding studyingand memorization. On a cumulative test, they would simply refer to their base of conceptualknowledge, which is not easily forgotten. Supporting this supposition, Bell (2002) foundthat males outperformed females on physical science questions that required the retrievalfrom memory of declarative knowledge. Perhaps this was because males were consultingbase knowledge rather than memory.

The last two variables in this section deal with the affective domain. The first is father’sencouragement to take science. The performance of females increased if they reportedthat their father encouraged them. Figure 5 illustrates the predicted shift in male andfemale university physics points when they report whether their father encouraged them.However, father’s encouragement had no significant effect on male performance. Nord,Brimhall, and West (1997, p. 56) found in a national study that children in grades 6–12“are more likely to get mostly A’s if their fathers are involved in their schools.” However,they did not analyze their data with respect to the gender of the student or specific subject


870 HAZARI ET AL.

Figure 5. Shift in university physics grade percent (GRDPRCNT) predicted by gender and father’s encouragementto take physics.

area achievement. Additional research has found that a father’s perception of his child’sphysical science ability relates to the child’s perception of his/her own ability (Andreet al., 1999). Specific to daughters, research indicates that having a demanding yet supportivefather, who relates to his daughter as he would to his son, seems to encourage mental-healthgrowth in females (Perkins, 2001; Secunda, 1992). In addition, Scott and Mallinckrodt’s(2005) study indicates that young women who have positive relationships (supportive andaffectionate, not controlling or conditionally affectionate) with their fathers are more likelyto pursue nontraditional careers such as becoming a scientist. Since physics continues tobe a male-dominated field, where men are the empowered group, the influence on femalesby their fathers (as a member of the empowered group) may be of particular interest. Thisobservation is supported by historical trends since many female physicists have reportedhigh levels of influence their fathers have had on their careers. Wertheim (1995), whorecounts in her book the history of females in physics, writes of the first-recorded femalescientist and mathematician in history, Hypatia,

In a scenario we shall see repeated many times in this book, her rare good fortune wasdue to the enlightened attitude of her father, Theon, a mathematician and astronomer whotaught her himself. Indeed, up until the twentieth century, almost all mathematical womenwere taught by a male relative, usually a father or husband. (p. 35)

In interviews with female scientists, Gilbert and Calvert (2003) found that “all showed ahigh degree of attunement towards their fathers, both in terms of their ‘thinking styles’,and in a deeper emotional sense” (p. 873). Thus, the notion that fathers may have astrong affective role in influencing females’ physics attitudes and performance is likely.This research supports the finding that father’s encouragement may influence femaleperformance in physics. Another possibility is that fathers might encourage their daughters



Figure 6. Shift in university physics grade percent (GRDPRCNT) predicted by gender and family belief in sciencefor a better career.

more selectively if they perceive a strong science aptitude in them.The second significant affective interaction involves a variable that positively influenced

male performance but had no significant impact on female performance. This variable isthe family attitude that science is “a way for you to have a better career.” As before, Figure6 illustrates the predicted shift in male and female students’ university physics pointswhen they report whether their family believed that science was for a better career. It isnot surprising that this attitude had a strong influence on male performance but none onfemale performance. Females generally put less emphasis on having a high status careerthan men do (Morgan, Isaac, & Sansone, 2001). In fact, females have been found to havemore interpersonal career goals (Morgan et al., 2001) and want to participate in sciencethat “helps other people” (Jones et al., 2000). Thus, the attitude that science is for a “bettercareer” would have little impact on their motivation to perform and succeed. Males, on theother hand, have reported liking and wanting to pursue science in order to control otherpeople, become famous, earn lots of money, and receive high status (Jones et al., 2000;Morgan et al., 2001). Thus, a family attitude that science is for a “better career” may serveto motivate males to perform and succeed.

CONCLUSIONS

Several issues can be raised from this work that are particularly useful in highlightingproblems with high school and university physics education. First, high school physicspedagogy that is generally believed to positively influence students’ future performancemay actually have little, no, or even negative impact on university performance the way it isbroadly being implemented. This type of epidemiological study using HLM can be powerfulbecause it brings to the foreground possible problems with pedagogical implementationin high school. The word “implementation” is used because it is not the pedagogy in


872 HAZARI ET AL.

itself that is good or bad but how and when it is used that makes it effective or not.For example, a laboratory exercise might be well designed and implemented such that itpositively influences learning and later university performance just as a laboratory exercisecan be poorly designed and implemented leaving students confused or having wasted timesuch that it negatively influences university performance. However, this leads to an obviouslimitation in our study in that we have no information on how the pedagogy was actuallybeing implemented. All we know is its broad ability to predict performance.

Second, if pedagogy that is valued by many science teachers and education researchers,such as independent projects, does not lead to better university performance, perhaps thisis an indication of fallacies within the university physics curriculum rather than with thepedagogy or high school implementation itself. In other words, perhaps the universityphysics curriculum is in need of serious instructional reform and traditional universityphysics education is not the standard we should be preparing students for in high school.Having said this, it is still important for university physics to not be an impediment tostudents seeking a career in science. Thus, high school teachers have the onus of strikingthe balance between preparing their students for success in a university course as wellas providing them with “good” physics instruction whether that instruction helps them inuniversity or not. This issue stems from the often conflicting goals that teachers have todeal with of preparing future science students for university science and preparing futurenonscience students for everyday life and a lifelong appreciation for physics. Perhaps, itis too much to ask teachers to juggle these goals without at least giving a critical lookto reforming university curriculum so that it does not punish alternate forms of “good”instruction such as a humanistic approach that connects students to the concerns of everydaylife and society (Aikenhead, 2005).

In addition, this study implies that when choosing pedagogy, gender often matters. Peda-gogy that works for one student may not work for another and traditional physics pedagogyhas historically catered to the male majority. Pedagogy and curriculum further have theability to reproduce the participants of a discipline by defining learning and knowledgethat favor a particular group. In other words, physics learning and knowledge cannot beseparated from the history, organization, composition, and social activity of the communityitself (Nespor, 1994). However, this makes changing the physics curriculum, especially atthe university level where we enter the realm of the physicist, increasingly complex.

Although this work does not imply causal relationships between the predictors andoutcome variable, it does provide insight into what may or may not be working in highschool in terms of preparing students for university physics. Thus, given the cited pastresearch that supports these results and provided future research also supports these results,some suggestions can be made to high school physics teachers as avenues to try out in orderto prepare students for university physics. These suggestions can also serve as researchquestions on pedagogy implementation that teachers and researchers can try to answer.Also, since the suggestions do not take into account the important role that the pedagogiesmight play in terms of attitude and interest, further research is also necessary within thatdomain. The suggestions are as follows:

• Spend more time on less content (more time on mechanics and optics and historyof physics a recurring topic); that is, concentrate on understanding and depth overbreadth.

• Increase frequency of physics-related videos but limit or reform implementation ofstudent-designed projects, reading and discussion of labs the day before performingthem, microcomputer-based labs, and postdemonstration discussions.

• Include problems requiring calculations on tests/quizzes.



The recommendations with regard to balancing gender differences (drawn from the inter-section area of the male and female graphs) are as follows:

• Have components that require students to have a deeper understanding of concepts.Although some memorization is acceptable, students should not be able to rely largelyon rote-learning strategies. One strategy might be to use many real-world examplessince real-world examples were highly correlated with conceptual understandinglearning requirements.

• Limit/reform long written problem usage and cumulative assessments (e.g., less“cramming,” more understanding).

Again, the results of this study indicate that there are pedagogies and affective factors thatmight influence male and female students differently. Perhaps, this is part of the reason thatsingle-sex education using reformed curricula and teachers trained to develop students’self-concept has been so successful in improving female performance and persistence(Gillibrand et al., 1999; Haussler & Hoffmann, 2002; Parker & Rennie, 2002). Thesereforms concentrate on the pedagogy and affective support that work for females. Single-sex classrooms are one place where the results reported in this study could be further testedthrough implementation within actual classrooms.

Finally, some may contend that helping females perform better in the “same old” physicscourses perpetuates bias by trying to fit females into the “correct” mold. This is oneimportant perspective. Another perspective is to recognize that similar to other disciplines,females are increasing in number and percentage in the field of physics as time progresses.If helping them perform and stay interested at each stage can facilitate their entry into thefield, then eventually they may be able to change the field from within as citizens of thefield. This may be a more realistic approach than trying to reinvent the field from outsideits borders. In a sense, this study is an attempt to facilitate the “border crossing” of femalesinto the world of physics so that they do not enter into the domain as a lower class, at least interms of performance. The next step is to ascertain high school physics and affective factorsthat influence female interest in physics as well as the factors that influence a female’schoice to pursue a physics or physics-related degree (e.g., engineering) in university.

The authors would like to thank A. Trenga, M. Filisky, J. Loehr, B. Ward, J. Peritz, F. Deutsch, C.Crockett, M. Schwartz, and other members of the FICSS team for their dedicated work. We alsothank the National Science Foundation (NSF) and all the physics professors and their students formaking this study possible. The primary author would like to thank G. Potvin, G. Sonnert, J. Miller,H. Coyle, D. Hodson, M. Nieswandt, E. Jang, E. Pedretti, J. Ebenezer, I. Decoito, A. Sharkawy, andK. Bellomo for helpful revisions/discussions, as well as the Social Sciences and Humanities ResearchCouncil of Canada (SSHRC) for their continued support.

REFERENCES

Aikenhead, G. (1996). Science education: Border crossing into the subculture of science. Studies in ScienceEducation, 27(1), 1 – 52.

Aikenhead, G. (2005). Science education for everyday life: Evidence-based practice. New York: Teachers CollegePress.

Allison, P. D. (2002). Missing data: Quantitative applications in the social sciences. Thousand Oaks, CA: Sage.Alsop, S., & Watts, M. (2003). Science education and affect. International Journal of Science Education, 25(9),

1043 – 1047.Alters, B. J. (1995). Counseling physics students: A research basis. The Physics Teacher, 33(6), 413 – 415.Andre, T., Whigham, M., Hendrickson, A., & Chambers, S. (1999). Competency beliefs, positive affect, and


874 HAZARI ET AL.

gender stereotypes of elementary students and their parents about science versus other school subjects. Journalof Research in Science Teaching, 36(6), 719 – 747.

Baker, D., & Leary, R. (2003). Letting girls speak out about science. Journal of Research in Science Teach-ing,40(Suppl.), S176 – S200.

Barker, B. (1997). Girls’ world or anxious times: What’s really happening at school in the gender war. EducationalReview, 49(3), 221 – 227.

Barron, B. J. S., Schwartz, D. L., Vye, N. J., Moore, A., Petrosino, A., Zech, L., et al. (1998). Doing withunderstanding: Lessons from research on problem and project-based learning. Journal of the Learning Sciences,7(3/4), 271 – 311.

Bell, J. (2001). Investigating gender differences in the science performance of 16-year-old pupils in the UK.International Journal of Science Education, 23(5), 469 – 486.

Bouchard, P., St-Amant, J. C., & Deslandes, R. (1998). Family variables as predictors of school achievement: Sexdifferences in Quebec adolescents. Canadian Journal of Education, 23(4), 390 – 404.

Brickhouse, N., Lowery, P., & Schultz, K. (2000). What kind of a girl does science? The construction of schoolscience identities. Journal of Research in Science Teaching, 37(5), 441 – 458.

Carlone, H. (2004). The cultural production of science in reform-based physics: Girls’ access, participation, andresistance. Journal of Research in Science Teaching, 41(4), 392 – 414.

Cavallo, A. M. L., Potter, W. H., & Rozman, M. (2004). Gender differences in learning constructs, shifts in learningconstructs, and their relationship to course achievement in a structured inquiry, yearlong college physics coursefor life science majors. School Science & Mathematics, 104(6), 288 – 300.

Chambers, S., & Andre, T. (1997). Gender, prior knowledge, interest, and experience in electricity and conceptualchange text manipulations in learning about direct current. Journal of Research in Science Teaching, 34(2),107 – 123.

Cleaves, A. (2005). The formation of science choices in secondary school. International Journal of ScienceEducation, 27(4), 471 – 486.

Conley, D. T. (2003). Understanding university success. Eugene, OR: Center for Educational Policy Research,University of Oregon.

Costa, V. (1995). When science is “another world”: Relationships between worlds of family, friends, school, andscience. Science Education, 79(3), 313 – 333.

Crouch, C. H., Fagen, A. P., Callan, J. P., & Mazur, E. (2004). Classroom demonstrations: Learning tools orentertainment? American Journal of Physics, 72(6), 835 – 838.

Dimitrov, D. (1999). Gender differences in science achievement: Differential effect of ability, response format,and strands of learning outcomes. School Science & Mathematics, 99(8), 445 – 451.

Epstein, D. (1998). Real boys don’t work: ‘Underachievement’, masculinity, and the harassment of ‘sissies’. InD. Epstein, J. Elwood, V. Hey, & J. Maw (Eds.), Failing boys? Issues in gender and achievement. Buckingham,England: Open University Press.

Forgione, P., Nohara, D., Welch, C., & Salganik, L. (2000). Secondary performance. The Science Teacher, 67(1),60.

Fratt, L. (2002). Less is more: Trimming the overstuffed curriculum [Electronic Version]. District Administration,38. Retrieved May 3, 2007, from http://districtadministration.ccsct.com//page.cfm?p=228.

Freeman, C. E. (2004). Trends in educational equity of girls & women: 2004. National Center for EducationStatistics Report. Retrieved May 3, 2007, from http://nces.ed.gov/pubsearch/pubsinfo.asp?pubid=2005016

Gainen, J. (1995). Barriers to success in quantitative gatekeeper courses. In J. Gainen & E. W. Williamsen (Eds.),Fostering student success in quantitative gateway courses (pp. 5 – 14). San Francisco, CA: Jossey-Bass.

Gifford, V. D., & Harpole, S. (1986). Factors contributing to freshmen physics achievement. Paper presented atthe Mid-South Educational Research Association, Memphis, TN.

Gilbert, J., & Calvert, S. (2003). Challenging accepted wisdom: Looking at the gender and science educationquestion through a different lens. International Journal of Science Education, 25(7), 861 – 878.

Gillibrand, E., Robinson, P., Brawn, R., & Osborn, A. (1999). Girls’ participation in physics in single sex classesin mixed schools in relation to confidence and achievement. International Journal of Science Education, 21(4),349 – 362.

Hart, G. E., & Cottle, P. D. (1993). Academic backgrounds and achievement in college physics. The PhysicsTeacher, 31(8), 470 – 475.

Harwood, W. S., & McMahon, M. M. (1997). Effects of integrated video media on student achievement andattitudes in high school chemistry. Journal of Research in Science Teaching, 34(6), 617 – 631.

Haussler, P., & Hoffmann, L. (2002). An intervention study to enhance girls’ interest, self-concept and achievementin physics classes. Journal of Research in Science Teaching, 39(9), 870 – 888.

Hazari, Z., & Potvin, G. (2005). Views on female under-representation in physics: Retraining women orreinventing physics? Electronic Journal of Science Education, 10(1), 1 – 33. Retrieved May 3, 2007, from



http://wolfweb.unr.edu/homepage/crowther/ejse/ejsev10n1.html.Hazel, E., Logan, P., & Gallagher, P. (1997). Equitable assessment of students in physics: Importance of gender

and language background. International Journal of Science Education, 19(4), 381 – 392.Heller, P., & Hollabaugh, M. (1992). Teaching problem solving through cooperative grouping, Part 2: Designing

problems and structuring groups. American Journal of Physics, 60(7), 637 – 644.Hughes, G. (2001). Exploring the availability of student scientist identities within curriculum discourse. Gender

& Education, 13(3), 275 – 290.Ivie, R., Cuzjko, R., & Stowe, K. (2001). Women physicists speak: The 2001 International Study of

Women in Physics. American Institute of Physics Report. Retrieved May 3, 2007, from http://www.aip.org/statistics/trends/gendertrends.html

Jackson, C. (2002). ‘Laddishness’ as a self-worth protection strategy. Gender & Education, 14(1), 37 – 51.Jones, G., Howe, A., & Rua, M. (2000). Gender differences in students’ experiences, interests, and attitudes

toward science and scientists. Science Education, 84(2), 180 – 192.Kenway, J., & Gough, A. (1998). Gender and science education in schools: A review “with attitude.” Studies in

Science Education, 31(1), 1 – 30.Kulik, J. A. (2002). School mathematics and science programs benefit from instructional technology. InfoBrief:

Science Resource Statistics. Retrieved May 3, 2007, from http://www.nsf.gov/statistics/infbrief/nsf03301/Labudde, P., Herzog, W., Neuenschwander, M., Violi, E., & Gerber, C. (2000). Girls and physics: Teaching and

learning strategies tested by classroom interventions in grade 11. International Journal of Science Education,22(2), 143 – 157.

Laws, P., Rosborough, P., & Poodry, F. (1999). Women’s responses to an activity-based introductory physicsprogram. American Journal of Physics, 67(7), S32 – S37.

Lederman, M. (2003). Gender/InEquity in science education: A response. Journal of Research in Science Teaching,40(6), 604 – 606.

Leonard, D. K., & Jiang, J. (1999). Gender bias and the college prediction of the SATs: A cry of despair. Researchin Higher Education, 40(4), 375 – 408.

Lie, S., & Bryhni, E. (1983). Girls and physics: Attitudes, experiences and underachievement. Paper presented atthe Second GASAT Conference, London.

Little, R. J. A., & Rubin, D. B. (2002). Statistical analysis with missing data. New York: Wiley.Long, D. D., McLaughlin, G. W., & Bloom, A. M. (1986). The influence of physics laboratories on student

performance in a lecture course. American Journal of Physics, 54(2), 122 – 125.Marshall, J. A., & Dorward, J. T. (2000). Inquiry experiences as a lecture supplement for preservice elementary

teachers and general education students. American Journal of Physics, 68(S1), S27 – S36.Matthews, M. R. (1994). Science teaching: The role of history and philosophy of science. New York: Routledge.McCullough, L. (2004). Gender, context, and physics assessment. Journal of International Women’s Studies, 5(4),

20 – 30.Monk, M., & Osborne, J. (1997). Placing the history and philosophy of science on the curriculum: A model for

the development of pedagogy. Science Education, 81(4), 405 – 424.Morgan, C., Isaac, J. D., & Sansone, C. (2001). The role of interest in understanding the career choices of female

and male college students. Sex Roles, 44(5/6), 295 – 320.Musheno, B. V., & Lawson, A. E. (1999). Effects of learning cycle and traditional text on comprehension of

science concepts by students at differing reasoning levels. Journal of Research in Science Teaching, 36(1),23 – 37.

Nespor, J. (1994). Knowledge in motion: Space, time, and curriculum in undergraduate physics and management.London: Falmer Press.

Nord, C. W., Brimhall, D., & West, J. (1997). Fathers’ involvement in their children’s schools. National Centerfor Educational Statistics Report. Retrieved May 3, 2007, from http://nces.ed.gov/pubs98/fathers

Parker, L., & Rennie, L. (2002). Teachers’ implementation of gender-inclusive instructional strategies in single-sexand mixed-sex science classrooms. International Journal of Science Education, 24(9), 881 – 897.

Perkins, R. (2001). The father – daughter relationship: Familial interactions that impact a daughter’s style of life.College Student Journal, 35(4), 616 – 626.

Rakow, S. J. (1999). NSTA’s response to the TIMSS. The Science Teacher, 66(1), 30 – 33.Redish, E. F., Saul, J. M., & Steinberg, R. N. (1997). On the effectiveness of active-engagement micro-computer

based laboratories. American Journal of Physics, 65(1), 45 – 54.Rivard, L. P., & Straw, S. B. (2000). The effect of talk and writing on learning science: An exploratory study.

Science Education, 84(5), 566 – 593.


876 HAZARI ET AL.

Roth, W. M., McRobbie, C. J., Lucas, K. B., & Boutonne, S. (1997). Why may students fail to learn fromdemonstrations? A social practice perspective on learning in physics. Journal of Research in Science Teaching,34(5), 509 – 533.

Roychoudhury, A., Tippins, D., & Nichols, S. (1995). Gender-inclusive science teaching: A feminist-constructivistapproach. Journal of Research in Science Teaching, 32(9), 897 – 924.

Rudge, D. W., & Howe, E. M. (2004). Incorporating history into the science classroom. The Science Teacher,71(9), 52 – 57.

Sadler, P. M., & Tai, R. H. (2001). Success in introductory college physics: The role of high school preparation.Science Education, 85(2), 111 – 136.

Scheffer, J. (2002). Dealing with missing data. Research Letters in the Information and Mathematical Sciences,3(1), 153 – 160.

Schneider, R. M., Krajcik, J., Marx, R. W., & Soloway, E. (2002). Performance of students in project-basedscience classrooms on a national measure of science achievement. Journal of Research in Science Teaching,39(5), 410 – 422.

Scott, A. B., & Mallinckrodt, B. (2005). Parental emotional support, science self-efficacy, and choice of sciencemajor in undergraduate women. The Career Development Quarterly,53(3), 263 – 273.

Secunda, V. (1992). Women and their fathers. New York: Bantam Doubleday Dell.Seymour, E. (2001). Tracking the processes of change in US undergraduate education in science, mathematics,

engineering, and technology. Science Education, 86(1), 79 – 105.Shumba, O., & Glass, L. W. (1994). Perceptions of coordinators of college freshman chemistry regarding selected

goals and outcomes of high school chemistry. Journal of Research in Science Teaching, 31(4), 381 – 392.Singer, S. R., Hilton, M. L., & Schweingruber, H. A. (Eds.). (2006). America’s lab report: Investigations in high

school science. Washington, DC: National Academies Press.Smith, F. M., & Hausafus, C. O. (1998). Relationship of family support and ethnic minority students’ achievement

in science and mathematics. Science Education, 82(1), 111 – 125.Snyder, T. D., Tan, A. G., & Hoffman, C. M. (2004). Digest of educational statistics, 2003. Washington, DC:

National Center for Educational Statistics.Sommers, C. H. (2000). The war against boys. The Atlantic Monthly, 285(5), 59 – 74.Stadler, H., Duit, R., & Benke, G. (2000). Do boys and girls understand physics differently? Physics Education,

35(6), 417 – 422.Tai, R. H., & Sadler, P. M. (2001). Gender differences in introductory undergraduate physics performance:

University physics versus college physics in the USA. International Journal of Science Education, 23(10),1017 – 1037.

Tai, R. H., Sadler, P. M., & Loehr, J. F. (2005). Factors influencing success in introductory college chemistry.Journal of Research in Science Teaching, 42(9), 987 – 1012.

Urry, M. (2003). Speeding up the long slow path to change. APS News, 12(2).Wainer, H., & Steinberg, L. S. (1992). Sex differences in performance on the math section of the SAT: A

bidirectional validity study. Harvard Educational Review, 62(3), 323 – 336.Wallsgrove, R. (1980). The masculine face of science. In L. Birke, W. Faulkner, D. Best, D. Janson-Smith, & K.

Overfield (Eds.), Alice through the microscope (pp. 228 – 240). London: Virago.Wertheim, M. (1995). Pythagoras’ trousers: God, physics, and the gender wars. New York: Times Books/Random

House.Williams, K. (2000/2001). Understanding, communication anxiety and gender in physics. Journal of College

Science Teaching, 30(4), 232 – 237.Xu, J. (2006). Gender and homework management reported by high school students. Educational Psychology,

26(1), 73 – 91.Zohar, A., & Sela, D. (2003). Her physics, his physics: Gender issues in Israeli advanced placement physics

classes. International Journal of Science Education, 25(2), 245 – 268.


Gender differences in introductory university physics performance: The influence of high school physics preparation and affective factors

Documents