An Exploratory Study on Recruitment and Retention of Women in Underg raduate CS: Collaborative Learning with Formal Learning Groups

By Julie Krause and Irene Polycarpou

Introduction

Percentages of women in computing jobs and in university programs are not where they should be. As a result, recruiting and retaining women in the field of Computer Science (CS) is of utmost importance. At our university, Colorado School of Mines (CSM), where only 13% of undergraduate students enrolled in the CS program are female, we are looking into ways to promote female participation in CS.

The use of collaborative learning in various forms has been shown to improve recruitment and retention of women in CS studies (Chase & Okie, 2000; Horwitz & Rodger, 2009; McDowell et al., 2006; NCWIT, 2011; NCWIT, Barker, & Cohoon, 2008). Since our Introduction to CS course (CSCI 101) at CSM incorporates formal learning groups (Johnson, Johnson & Smith, 1991), which is a form of collaborative learning, we decided to conduct an exploratory study on formal learning groups’ potential to promote recruitment and retention of women in our CS program. Our work was supported by a National Center for Women & Information Technology (NCWIT) Academic Alliance Seed Fund award sponsored by Microsoft Research. In this article, we discuss the findings of our exploratory study.

About Formal Learning Groups in Introduction to CS

In CSCI 101, the introductory CS course at CSM that was first offered during the Fall 2010 semester, students work in formal learning groups as an integral part of their learning experience. Teams of four to five students work together in class over a period of several weeks, and each student participates in approximately four different groups throughout the semester. With the exception of the first group formation of the semester, which is completely random, group formations are determined using a randomised algorithm that, based on students’ previous grades (including programming assignments and quiz scores), creates groups with similar average performance. As a result, groups have varying combinations of female and male students, as well as CS majors and non-majors.

Much of the content of CSCI 101, which is a course designed to present a broad variety of CS concepts, is covered via formal learning groups. For at least twenty minutes of each fifty-minute class, students work in groups to do problem-solving as a team or to teach one another about topics they have investigated on their own time. During each class period, students are presented with a “learning group assignment” and are given time to divide up problems among group members. Each group member is then responsible for investigating specific topics and completing related problems outside of class. Groups are encouraged to duplicate coverage on completion of problems between two or more students. In the subsequent class, students teach one another what they have learned. If different students come up with different solutions, they can do additional work as a group to resolve discrepancies and ask the instructor for additional clarification.

Throughout group discussion time in class, instructors observe groups and answer content-related questions. Furthermore, during this time, instructors verify that students have come to the group sessions prepared. Students’ course grades are negatively impacted if they come unprepared numerous times throughout the semester. Following the group discussion time, an instructor facilitates a class-wide question and answer session to clarify content questions, delivers a short lecture, and often assigns problem-solving tasks to groups.

As motivation for students to be invested in teaching their peers, 10% of each student’s course grade is dependent on their group members’ performance on individually completed quizzes, which are administered at the end of the multi-week period that a given set of groups worked together.

For the course, when the class size is 40-60 students there are two instructors – as was done during the semester of our study. When the class size is fewer than 40 students, there is only one instructor.

Study Methods

We used both quantitative and qualitative data collection and analysis for our study. Our research included collecting data from students, so we obtained approval from our Institutional Review Board prior to beginning our research. We conducted surveys with open-ended and closed-option questions including surveys administered at the beginning of the semester and at the end of the course. Additionally, one-on-one interviews and focus groups were conducted at the end of the semester to gain more in-depth insights into students’ perceptions. The participants of the study were students who were enrolled in CSCI 101 during the Spring 2011 semester. All students (20 females and 82 males) were invited, but not required, to participate. Fifteen female students and 61 male students (76 students total) responded to all of the surveys. Additionally, a randomly selected subset of students was invited to participate in an interview or focus group. We invited all female students, 12 of whom agreed to participate. Whether they were selected for an interview or a focus group was random. The number of male students selected was chosen to match the number of female participants. Four female students and four male students participated in interviews. We conducted two female focus groups with four students each and two male focus groups with three students each; the discrepancy was due to the fact that two male students did not attend their focus groups.

The surveys, interview questions, and focus group guides were designed to garner students’ perceptions of their experiences with learning groups in CSCI 101. We also measured changes from the beginning to the end of the semester in students’ intent to pursue CS studies to determine if recruitment and retention were successful. Furthermore, we investigated different group formations (e.g., combinations of female/male students) to explore whether certain formations were more or less beneficial to female students, and to identify properties of groups that students associate with more effective groups.

Findings

Recruitment and Retention Success

We found that there were no significant changes in students’ interest in majoring in CS after taking CSCI 101. Of the students who came into the course as CS majors, all but one male student retained an interest in the major, indicating that retention of students in the major was primarily positive. Of the students who came into the course undeclared or with non-CS majors (roughly three-quarters of the female students and one-half of the male students), two male students lost interest in majoring in CS, whereas one female and two male students gained an interest in majoring in CS. While there was one female student who decided after taking CSCI 101 that she was then interested in pursuing a CS major, overall, these results indicate that recruitment into the major was not necessarily successful.

Students’ Perceptions of Formal Learning Groups

Through surveys, students reported on how they compared learning groups to traditional learning methods and whether learning groups had an impact on their intent to study CS. Through interviews, focus groups and open-ended survey questions, students also reported on their general perceptions of learning groups in CSCI 101.

Many female students who took the course reported that they learned from and enjoyed formal learning groups. For example, when asked to compare learning groups with traditional learning methods, more female students indicated that learning groups were positive compared to traditional methods than those who indicated that learning groups were negative compared to traditional methods. Male students, on the other hand, indicated a preference for traditional methods. When students were asked to rank five instructional methods used in CSCI 101 (participating in learning groups, reading the textbook, observing lecture, programming in Python, and playing educational games in class) in terms of how much they learned from each method, many female students (40% of the females who took the surveys) ranked learning groups as the method from which they learned the most. Similarly, 40% of the female students ranked learning groups as the method they enjoyed most.

When asked if learning groups had an impact on their intent to study CS, most students indicated that learning groups did not have any impact. However, of the two female students who indicated that learning groups did have an impact on their intent to study CS, both students indicated a positive impact.

Some of the most interesting findings were the specific benefits and drawbacks of learning groups that students reported in open-ended survey responses, in interviews, or in focus groups. Positive aspects of learning groups that students reported were that having group members depend on them provided additional motivation to work hard, that learning groups helped them to feel comfortable asking questions due to the small group setting (as contrasted with asking questions in front of a large class), and that learning groups gave students the opportunity to meet people. The most common benefits that female students in particular reported were the added comfort in participating and the additional interactivity with others that learning groups provided.

Some of the concerns about learning groups that students reported were that working with groups had the potential to cause misconceptions or poor understanding of content, and that the quality of the learning group experience varied greatly depending on the group members (e.g., some groups were open to teaching, whereas others did not teach sufficiently and students felt they needed to learn everything on their own as a result). Although learning group discussions in CSCI 101 were monitored and instructors provided clarifications regarding content misconceptions as needed, these findings indicate that a rigorous approach to observing and influencing learning group interactions is needed for learning groups to be most successful. For example, instructors could observe groups and assign grades for students’ participation during the observation.

Specific Group Formations

Students completed surveys to evaluate their specific groups, and those survey data were analyzed to determine whether specific group formations had an impact on female students’ experiences in learning groups within CSCI 101.

One surprising outcome was that we found that female students in our study did not necessarily prefer working in groups with other female students (i.e., groups consisting of more than one female). Although many of the groups that were reported by female students to be successful in general were comprised of more than one female, when female students were asked directly about their preferences, there were a variety of responses. We acknowledge that our sample size was small – 11 female students responded to this question during an interview or focus group. Five women reported that they had no preference – for example, some stated that they were used to working with all males, so it did not make a difference. Three women reported that they did prefer having other females in their groups, stating reasons such as not feeling ignored in those groups. Finally, three women reported that they preferred being the only female in a group. The students who indicated that they prefer groups without other females reported reasons such as getting along better with male students and not needing to compare oneself to other female students when there are no other females in the group.

There was also a spread in students’ preferences for homogeneity or heterogeneity in levels of CS experience and ability levels within groups, although there was a slight tendency toward a preference for having a mixture. Students’ comments suggest that there may be advantages in constructing learning groups in CSCI 101 with varying experience levels, but that groups with extreme variations are only beneficial if the highly experienced students are not condescending to the less experienced students, but rather take on the role of teacher in the group.

Students were given the opportunity in a survey to provide details regarding why their best groups were more effective or more inclusive, and why their worst groups were less effective or less inclusive. The characteristics of effective groups that appeared to be most important to students were preparedness (group members coming into group discussions prepared to teach their assigned content), motivation, and willingness to contribute to the learning group discussions (especially providing thorough explanations). For students to feel included in the teaching and learning process within groups, it was important for the groups to socialize and have fun together, to have members who were friendly, and to have members who listened to and valued other students’ input.

While the properties that students described were not necessarily properties that can be controlled by placing specific combinations of students together in groups, students’ feedback may be applied to improving learning groups. For example, to promote inclusive behaviors in group members and help students to feel that their input is valued, it may be beneficial for instructors to collect and act upon feedback regarding other group members’ performance to encourage positive behaviors and discourage negative ones.

Concluding Remarks

Results from our study, indicating that female students enjoyed and learned from formal learning groups in CSCI 101, suggest that incorporating learning groups into an introductory CS course may have the potential to promote female students’ interest in CS. There were not any specific combinations of students that were identified as being decidedly more or less beneficial to female students’ perceptions of groups, but certain characteristics of groups, such as motivation, preparedness, and approachability, appear to be needed for fostering positive group environments.

CSM continues to offer CSCI 101 with formal learning groups and will continue to hone the implementation to optimize learning for all students and to attract more women into CS. We hope that others may be inspired to use collaborative learning in their classrooms as a way to boost female students’ confidence in participating and hopefully to encourage their interest in CS. We also hope others will take away some valuable lessons from our findings and experiences.

About the Authors

Julie is a recent CSM M.S. graduate who worked on this study as part of her Master’s thesis under the advisement of Dr Irene Polycarpou. Julie was motivated to work on this project while co-teaching CSCI 101.

References

Chase, J. D., & Okie, E. G. (2000). Combining cooperative learning and peer instruction in introductory computer science. SIGCSE’00. ACM, New York, NY, 372-376.

Horwitz, S., & Rodger, S. H. (2009). Using peer-led team learning to increase participation and success of underrepresented groups in introductory computer science. SIGCSE’09, ACM, New York, NY, 163-167.

Johnson, D., Johnson, R., & Smith, K. (1991). Active learning: Cooperation in the college classroom. Edina, MN: Interaction Book Company.

McDowell, C., Werner, L., Bullock, H. E., & Fernald, J. (2006). Pair programming improves student retention, confidence and program quality. Commun. ACM. 49, 8 (Aug. 2006), 90-95.

NCWIT (2011). Promising Practice Catalog.

NCWIT, Barker, L., & Cohoon, J.M. (2008). Promising Practices Sheet: Retaining Women through Collaborative Learning.

---