Preliminary evidence for available roles in mixed-gender and all-women lab groups
PPreliminary evidence for available roles in mixed-gender and all-women lab groups
N.G. Holmes and Z. Yasemin Kalender
Laboratory of Atomic and Solid State Physics, Department of Physics, Cornell University, Ithaca, NY, 14850
Group work during lab instruction can be a source of inequity between male and female students. In this preliminarystudy, we explored the activities male and female students take on during a lab session at a university in Denmark.Different from many studies, the class was majority-female, so three of the seven groups were all female and the restwere mixed-gender. We found that students in mixed-gender groups divide tasks in similar ways to mixed-gender groupsat North American institutions, with men handling the equipment and women handling the computer more often. We alsofound that women in single-gender groups took on each of the available roles with approximately equal frequency, butwomen in single-gender groups spent more time on the equipment than students in mixed-gender groups. We interpretthe results through poststructual gender theory and the notion of ‘doing physics’ and ‘doing gender’ in physics labs. a r X i v : . [ phy s i c s . e d - ph ] J u l . INTRODUCTION AND MOTIVATION Collaborative group work is pervasive in modern physicseducation. Laboratory (lab) instruction typically employsgroup work, where group members must coordinate a diversearray of tasks to accomplish a common goal. Previous workhas found that male and female students (on average) dividesuch tasks inequitably [1–5]. In some cases, male studentspredominantly take on equipment manipulation [3–5] or dataanalysis [2], while female students predominantly take onmanagerial roles [1, 4] or computer work [3]. Recent evi-dence suggests that men take on different roles depending onwhether they are in single- or mixed-gender groups [3].Researchers are exploring multiple avenues to understandthe nature of or mechanisms for these differential roles. Inthis work, we are inspired by the theoretical framework, de-veloped by Danielsson and colleagues, that explores students‘doing gender’ and ‘doing physics’ in the context of labora-tory work [4, 6]. This framework builds from post-structuralgender theory in that gender is “created and negotiated by theindividual in response to a specific social setting” [6, p.27]and it is performed as individuals take on positions availableto them [7]. In physics lab activities, students can take onmultiple positions or roles, such as handling the equipment,analyzing data, or documenting the activities. Because stu-dents typically work in small groups to complete their exper-iments, these tasks understandably get divided between thegroup members. Thus, as students negotiate ‘doing physics’during their investigations, the social setting requires them toalso negotiate ‘doing gender’.With this lens, we interpret the emergence of distinct rolesfor male and female students in the previous literature as stu-dents’ enacting of gender in the laboratory. For example, menhandling the equipment more [3, 5] may relate to perform-ing masculinities such as tinkering [1, 4, 6, 8, 9]. Womenhandling the computer [3] or engaging in other non-technicaltasks [2] may relate to performing feminities such as commu-nication, secretary, or manager roles [1, 8, 9].While many of the observations of gendered division oftasks in physics labs took place in North America, the theoret-ical framework was developed by European researchers. Casestudies across the United States, Canada, and Sweden, how-ever, found remarkable similarities between the emergence ofgender in doing physics and lab work [9]. We sought to ex-tend this international focus to evaluate whether the gendereddivision of tasks previously observed at North American in-stitutions [1–3] would also emerge at a European institution.Furthermore, the theoretical framework related to ‘doinggender’ and ‘doing physics’ was linked to associations be-tween physics and masculinity, with women as visible, nu-merical minorities in the classroom or discipline [4, 9]. Sim-ilarly, the observations of gendered divisions of tasks tookplace in majority-male classrooms [1–3]. Relatively littlework has evaluated task divisions in female-majority class-rooms and in all-women groups.To pursue these questions, we conducted a preliminary, ob- servational study of students at an institution in Denmark dur-ing a single lab session where women were the numerical ma-jority. We aimed to shed light on two questions:1. Do male and female students in mixed-gender groupsin Denmark exhibit similar inequitable divisions oftasks as those observed at North American institutions?2. In a female-majority classroom, how do students dividetasks in all-women versus mixed-gender groups?While the gender dynamics in Denmark are socially andpolitically different than in the United States, the work byDanielsson with students in Sweden [4, 6, 8, 9] led us topredict that the observations of gendered tasks may be sim-ilar to those in North America. We had no predictions forthe behavior of women in single-gender groups, because thestudies exploring student roles in labs described above hadobserved majority-male classrooms and focused on mixed-gender groups.
II. METHODS
In this preliminary study, we measured coarse-grain behav-iors of students in an introductory mechanics course at a tech-nical university in Denmark (majority engineering majors inthe course and the institution).
A. Participants and instructional context
The data were collected from observations of a single labsession. The observed lab session was the first and only labsession for the course. The course traditionally did not in-volve any lab activities, but the instructors were in the pro-cess of developing a new laboratory curriculum to accom-pany the course. The new lab curriculum aims to developstudents’ experimentation and critical thinking skills, whilealso supporting the conceptual physics relevant to the lecture.The observed lab session was a pilot session for the new labcurriculum and students volunteered to participate (and weregiven pizza as incentive). The two-hour lab session was aboutprojectile motion. Students were tasked with collecting datato determine the launcher’s initial velocity and then to use thatinformation, combined with concepts and equations from lec-ture, to hit an arbitrary target set by the instructors.Of 50 students enrolled in the course, 21 students volun-teered to participate in the optional lab session. Althoughthe course enrolled approximately equal numbers of men andwomen, the students in the lab session were predominantlyfemale (16 women and 5 men). All students were phys-ical science or engineering majors from a variety of sub-disciplines. This was their first college-level physics course,though most students completed high school-level physics.Students self-selected into groups of two to four students.Three instructors and two graduate teaching assistants at-tended the lab session to facilitate and observe. Two re-searchers were also present to observe and collect data, butdid not interact with the students pedagogically.1 . Data collection
The two researchers stood at the back of the lab room forthe duration of the session. One of the observers took qual-itative field notes of the students’ activities. All instructionand most conversations took place in Danish, and neither re-searcher spoke Danish. Instructors periodically checked inwith the researchers to provide context for the conversationsthat could not be interpreted based on the visual activities.The second researcher took structured observations of stu-dent activities based on a protocol used in Refs. [2, 3]. Theprotocol has demonstrated strong inter-rater reliability [3] andthe observer in this study had previously used this protocolwith sufficient interrater reliability (citation omitted for con-fidential review). With this protocol, the researcher recorded,in two to three minute intervals, whether each student hadtheir hands on the experimental equipment, a computer orlaptop, or pencil and paper, or whether they were talking totheir group, another group, or the instructor. Any other be-haviors were coded as ‘Other.’ These activities have been pre-viously compared to qualitative descriptions of the students’experimentation tasks, such that we can approximately inferexperimentation roles from their tasks [3]. That is, handlingthe equipment reflects either setting up (or taking down) theapparatus or collecting data, handling the computer reflectsentering data, analyzing the data, or documenting methods inlab notes, handling paper reflects making notes or consultingthe lab manual, and so on. The data indicated that very fewstudents interacted with other group members, so this codewas collapsed with ‘Other.’ If the student was shifting activi-ties, the observer waited no more than 10 seconds to identifythe dominant activity. With 21 students present, each stu-dent’s behavior could be documented within the two to threeminute window, moving sequentially around the room. A to-tal of 43 intervals were coded for each student.
C. Analysis
The coding protocol provides coarse-grain, efficient, andbroad information about all students in the class. The proto-col does not attend to fine-grain interactions or student con-versation. From previous work with the protocol, the cod-ing categories effectively estimate and distinguish forms ofstudent engagement in various lab activities (e.g., analyzingdata is reserved to handling the computer, considering the ex-perimental set up or collecting data are reserved to handlingthe equipment, taking notes or consulting the lab manual arereserved to handling paper) [3]. The coarseness of the dataplaces limitations on the interpretations, but allows a broadsample of all students over a long period of time, providingan estimate of students’ roles and activities in the lab.We calculated the total number of coded intervals for eachstudent for each activity. We then produced scatter plots ofthe frequencies of intervals in which students were coded asengaging with each type of activity. We also explored the total number of coded intervals by each student for each activity,comparing within and between groups. We summarize thesedata into two figures. The first looks across all groups to iden-tify overall differences between men and women in mixed-and single-gender groups, including the mean and standarderror of the number of coded intervals across demographics.The second looks within each group to compare the behav-iors of each group member compared to their peers. Giventhe small sample size, statistical comparisons would not bemeaningful, so we rely only on the graphical analyses.Based on previous cluster analyses [3], we used the totalfrequencies across the lab period to estimate the students’dominant roles by evaluating each student’s engagement inthe activity compared to the rest of the class. That is, weconverted each student’s total frequency for each activity to a z -score relative to the rest of the class. For each person foreach activity, a z -score is calculated as: z = F i − Fσ (1)where F i is the total frequency for the individual for the givenactivity, and F and σ are the average and standard deviationof the class’ total frequency for the given activity, respec-tively. The z -score, therefore, indicates by how many stan-dard deviations the individual’s frequency differs from theclass average. A z -score greater than zero indicates that thestudent performed the activity more than average, while a z -score less than zero indicates that the student performed theactivity less than average. A student’s dominant activity wasidentified as the activity with their largest z -score comparedto the rest of the class. This method takes into account that,on average, students spent unequal amounts of time on differ-ent tasks. We then compared the proportion of students witheach type of dominant activity by group type and gender. III. RESULTS
Figure 1 shows the distribution of the number of stu-dents coded as engaging in each type of activity for vari-ous numbers of intervals, broken out by gender and grouptype (mixed- versus single-gender groups). Within mixed-gender groups, we see some evidence of men and womentaking on different tasks. For example, on average, men spentmore time on the equipment than women and women spentmore time on paper than men. Women’s use of the computerwas highly variable, with some women almost never usingthe computer and some almost exclusively using the com-puter. Discussing with the instructor, discussing with theirgroup, and participating in Other activities were similar be-tween men and women in mixed-gender groups. These re-sults appear similar to those observed previously [3].Comparing the two panels, we can identify some interest-ing contrasts between students in mixed- and single-gendergroups. The figure shows that women in single-gender groupsspent more time on the equipment than women in mixed-gender groups and, on average, spend more time on the equip-2IG. 1: Scatter and box plot of the number of coded intervals for each activity ( E quipment, C omputer, P aper, I nstructor, G roup, and O ther) for each student by gender and group type. The box plots give the median (thick central line) and the firstand third quartiles, and the whiskers extend across 1.5 times the interquartile range. The box plots for men on paper,discussing with the instructor, and doing other are too small to be visible.ment than they do on any of the other tasks. Three students inmixed-gender groups seemed to spend almost no time on theequipment, while no students in single-gender groups spentthat little time on the equipment. Time spent on the com-puter is much less variable than for women in mixed-gendergroups and the average is again more comparable to the useby men. Other than one student, women in single-gendergroups seemed to spend much less time discussing with theirgroup than students in mixed-gender group. Time spent onpaper or doing Other is quite similar between students in thetwo types of groups.Figure 2 provides context for each individual’s behaviorrelative to their group members. The figure presents the num-ber of intervals for which each student was coded engagingin each activity. We have colored each student so that theycan be identified for each task. We use this representationpredominantly to evaluate whether students shared tasks ordivided tasks. We identify groups that shared tasks (MG1,MG2, SG1, and SG2) as ones whose points are generallyclustered together. That is, all group members performed anactivity for similar numbers of intervals. We identify groupsthat divided tasks (SG3 and MG3) as ones whose points arequite spread out. That is, group members performed an ac-tivity for a different number of intervals, with some studentsparticularly high and others particularly low. MG4 generallyshare tasks, except for one female student who almost exclu-sively uses the computer and does little else. Overall, sharingor dividing tasks does not seem to depend on group composi-tion (i.e., single- versus mixed-gender groups).The results are further illuminated by the representation inFigure 3, which shows the number of students that were iden-tified as dominant in each type of activity, broken down bygender and group type.A student’s dominant role was iden-tified as the activity the student performs most above aver- age, compared to the rest of the class. We see that womenin single-gender groups were represented across all activityroles, while women in mixed-gender groups were dominantonly on computer, paper, or discussing with their group. Menwere never dominant computer or paper users. IV. DISCUSSION
We conducted a preliminary, observational study to under-stand how male and female students at an institution in Den-mark divide tasks during a lab. In line with existing litera-ture [9], the gender patterns of task allocation were similar toresults found in North America [1, 3]. That is, we find that, inmixed-gender groups, men spent more time handling equip-ment than women, and that women were dominantly engagedin working on the computer, discussing with the group, orhandling paper worksheets.We uniquely find that women in single-gender groups spentmore time handling the equipment than students in mixed-gender groups. Furthermore, women in single-gender groupswere equally likely to dominate any type of activity, withdominant tasks being defined relative to the whole class. Inthe mixed-gender groups, in contrast, men and women eachhad roles not available to them. For example, no women inmixed-gender groups were dominant on the equipment, talk-ing to the instructor, or doing other tasks and no men weredominant computer or paper users. We interpret this resultthrough our theoretical framework: in mixed-gender groups,students do not take on particular roles because those roles arenot available them. The lack of available roles for men andwomen in mixed-gender groups supports the notion that stu-dents were navigating both ‘doing physics’ and ‘doing gen-der’ [4, 6]. That is, while all of these roles are necessary fordoing the experiment, some are perceived as more available3IG. 2: Scatter plot of the number of intervals with which each student (shown as a dot) was coded as engaging in each type ofbehavior. The jitter function was applied so that individual dots did not sit on top of each other. Dots close together mean thatall group members spent similar amounts of time doing an activity, while spread suggests tasks were divided. Groups labeled‘SG’ represent single-gender groups and those labeled ‘MG’ represented mixed-gender groups.FIG. 3: Histogram of the number of students identified asdominant in each type of activity relative to the class, brokendown by gender and group type.to women than men and vice versa. Equipment-handling inparticular is said to be “a doing of a particular classed mas-culinity” [8, p.488]. Our evidence supports this claim in thatno women in mixed-gender groups were dominant equipmentusers. However, our data also suggest that these tasks are notinherently gendered. If we ignore gender in Fig. 3, studentswere similarly dominant across the activities in each grouptype. As a reminder, dominant activities are defined relativeto all students in the class, not within groups. We infer, withsupport from the framework, that all tasks were available tothe women in single gender groups [7] and that these stu-dents were not negotiating ‘doing gender’ [4, 6], but weresimply negotiating ‘doing physics’ as they each take up dif-ferent tasks. Perhaps surprisingly, however, we do not see that studentsin either group type necessarily divide the work equally. Thatis, some groups seemed to divide-and-conquer while othersseemed to share the work equally, with no patterns betweensingle- and mixed-gender groups. Thus, women in single-gender groups are still apt to take on roles where one or twostudents dominate, for example, the equipment (e.g., groupSG3). This raises questions about whether group composi-tion (based on gender) predicts the equity of the group, assuggested by previous literature [10–12].Because the data are limited in multiple ways, all resultsshould be considered tentative. We highlight, in particular,that a higher proportion of female students attended the labthan were enrolled in the course, which suggests that the stu-dents may not be representative of the general course popu-lation. However, the population provided a unique samplingopportunity to quickly and roughly probe hypotheses fromexisting literature, and thus warrant future study. In additionto collecting more data, future work should also explore howstudents navigate into these roles. While previous work foundthat most task allocations were not overt [3], analysis of stu-dent positioning and body language in the labs may shed lighton how roles are being assigned or assumed.
ACKNOWLEDGMENTS
This work was supported by a Denmark Technical Uni-versity Grant for the Development of Teaching Quality andthrough the Cornell University Active Learning Intitiative.We would like to sincerely thank Kristoffer Haldrup, CarstenKnudsen, and Ole Trinhammer for their leadership on the labpracticals at DTU.4
1] D. Doucette, R. Clark, and C. Singh, European Journal ofPhysics , 035702 (2020).[2] J. Day, J. B. Stang, N. G. Holmes, D. Khumar, and D. A. Bonn,Physical Review Physics Education Research , 020104(2016).[3] K. N. Quinn, M. M. Kelley, K. L. McGill, E. M. Smith,Z. Whipps, and N. Holmes, Physical Review Physics Educa-tion Research , 010129 (2020), ISSN 2469-9896, URL https://link.aps.org/doi/10.1103/PhysRevPhysEducRes.16.010129.[4] A. T. Danielsson and C. Linder, Gender and Education , 129(2009).[5] J. Jovanovic and S. S. King, American Educational ResearchJournal , 477 (1998).[6] A. T. Danielsson, Gender and Education , 25 (2012).[7] E. Honan, M. Knobel, C. Baker, and B. Davies, QualitativeInquiry , 9 (2000).[8] A. T. Danielsson, Cultural Studies of Science Education , 477 (2014).[9] A. J. Gonsalves, A. T. Danielsson, and H. Pattersson, PhysicalReview Physics Education Research , 020120 (2016).[10] P. Heller and M. Hollabaugh, American Journalof Physics , 627 (1992), ISSN 00029505, URLhttp://link.aip.org/link/?AJP/60/627/1{&}Agg=doihttp://ajp.aapt.org/resource/1/ajpias/v60/i7/p627{_}s1.[11] K. D. Tanner, Cell Biology Education