Methods


Our intention is to understand the details and complexities of undergraduate computer science students' experiences, beliefs, and attitudes about computing, and the nature of the educational institutions responsible for delivering a computer science education, so that interventions to close the gender gap in the field are more effective. Nothing about this work has been straight-forward. Issues of confidence, gender socialization, motivation, culture, play a part along with the history of computers, curriculum and pedagogy, and the specific nature of programming itself. We concur with Eccles's observation:

"With increasing data about gender differences in occupational choices at different points of educational process, research models have become increasingly complex, linking achievement-related beliefs, outcomes and goals to interpretive systems such as causal attribution, the input of socializers and gender role beliefs, as well as self-perception and the subjective value of the science/math/engineering task itself" (Eccles, 1994).

We believe that in order to understand the present we must understand the past. Our primary source of data are interviews with computer science students themselves. The opening question of our interview is "Can you tell me the story of you and computers?" We have elicited students' accounts of their first experiences with computers, who introduced them, who worked with them, and what did they like to do with the computer. We are interested in the issues students consider when they decide to major (or not major) in computer science. We have gathered accounts from each student about this decision. And then we are interested in what happens to students once they are studying computer science at the undergraduate level.

Our study is longitudinal, following students' experiences for two to four years. The heart of our data is transcripts of semi-structured, open-ended interviews with students which occur once per semester. Additional data is gathered through surveys, classroom observations, interviews with faculty, a small journal writing project, monitoring of electronic communication and forums.

Participants

There are two sets of participants: computer science majors, and non-computer science students who are enrolled in an introductory programming course. To date (April 15, 1998) we have conducted interviews with 51 female and 46 male computer science majors, and with 25 female and 5 male students who are not computer science majors, for a total of 233 interviews with 127 students.

Our sample consists of the majority of female CS majors at Carnegie Mellon, and a comparable sample of male majors. Among the 51 CS women in our sample are 24 European Americans, 16 international students, 8 Asian Americans and 3 African Americans. Among the 46 men are 28 European Americans, 7 international students, 6 African Americans and 5 Hispanics. We are concerned with the low representation of underrepresented minority groups in the program, which generally tracks the university average of 7-8%. We are also concerned with their experiences: on average, only about half of our African American and Hispanic students persist through graduation. While we have little data to work with, due to the low numbers of minority students, we have been interviewing them as part of our research, with the goal of using the insight gained in future projects.

Interviews

The interviews are semi-structured and open-ended, designed to elicit students' own experiences rather than their abstract thoughts. Each student is interviewed once per semester. The opening question of the first interview is "Can you tell me the story of you and computer science?" It is here where we hear how much experience they have, who introduced them to the field, what engaged them, and how they fit computing into the rest of their lives. At this opening interview we also ask them about why they decided to major in computer science. Subsequent interviews focus on the experiences they are having with their course work, peers, faculty, culture of the field; their focus of interest in computer science; their sense of belonging (or not) in the field; their aspirations. All interviews are conducted by either Jane Margolis or Faye Miller and last approximately one hour.

The interview guides are being continuously added to and revised as we find that certain issues require further clarification. For example, we have found it nearly impossible to give relative weight to different detachment factors; factors frequently shift and appear enmeshed with one another. Therefore, we have added questions that elicit students' own ranking of which issues they believe to be the most influential in their attachment and/or detachment from computer science.

Data Analysis

Narrative Summaries

We attempt to keep the participants' stories as whole as possible, to avoid "context stripping." We attempt to establish a full portrait of each student. Immediately after each interview, the interviewer writes a narrative summary of the interview. We rely heavily on narrative summaries of the interviews to ensure that we consider the interview excerpts in their full context. We have worked very hard at negotiating the tension between presenting our data as full portraits and the almost necessary "fracturing " of the data into discrete elements so that we can detect patterns across groups and categories (see Maxwell, p. 63).

Interviewing students multiple times allows us the opportunity to check and re-check our interpretations of each students' experiences. The Carnegie Mellon student custom of personal web pages gives us another opportunity to check whether what we conclude from their interviews resonates with what they publicize about themselves. We also follow the department's electronic discussion groups, and plan to hold focus groups about our working papers.

Coding

Interviews are audiotaped, transcribed and entered into the computer software NUDIST. The interview texts are coded for all issues that bear on students' attachment to and detachment from computer science. Codes have been developed on the basis of what students discuss, as well as issues we believe to be salient, based on our prior knowledge and theoretical hunches. Regularly, we make sweeps through the data, coding for issues that emerge as most relevant. For example, when we note how connected confidence and interest are for many female students, we code for this specific link throughout all of the interviews in our sample. This means that we go into our data numerous times. We search for patterns among codes, pay attention to intensity and emphasis of various issues for each student, and use code frequencies to help establish prevalence and relative importance of mutiple factors.

We are continuously aware of the challenges posed by this type of data analysis. How we, as the researchers, hear the interview itself, how we as listeners and readers read the transcripts, what we decide to hear, what we regard as important all affect our final analysis. We have found that if we read the interviews as male vs. female we hear a clear contrasting sound. But, if we read women's interviews against each other we hear more complexity, with the stories and motivations of the persisters sounding different from those who decide to leave.

A Challenge: Longitudinal Instability

Our subjects--being the live creatures they are--change and transform in response to their own inner workings and to their environment. We have learned how a student's interview can be so different from her/his preceding one. For example, those students who decide to transfer out often go through a period shortly before they decide to leave when they tell us they are very glad with their decision to have majored, and are planning to stick with it. They often appear to have resolved some of the key detachment issues. But, then in the following semester they decide to leave. We are tracking the cycles that students go through, and our multiple interviews can help us pinpoint critical decision-making points and factors.

Interdisciplinary Collaboration

At the 1997 Grace Hopper Celebration of Women in Computing, founder Anita Borg called for more collaborative research between computer science and women's studies. Our project turns out to be one of a very few projects based on this interdisciplinary model. We personally have found this insider/outsider (who is the insider and who is the outsider?) interdisciplinary collaboration between computer science and education/women's studies to be intellectually rewarding and challenging.

We have benefited greatly from many previous research studies of women in math, science and engineering, and have found the Seymour and Hewitt book Talking about Leaving (1997), which explores undergraduate reasons for leaving the sciences, to be especially helpful and provocative in generating hypotheses. We are constantly reflecting on their findings; we see ourselves as building on their work, with an opportunity to take a more in-depth look into one discipline. We are also provoked by the ideas and writings of Carol Gilligan, Sue Rosser, Jo Sanders, Janet Schofield, Claude Steele, Sheila Tobias, Sherry Turkle, and other researchers in this field.

References

Eccles, J. (1994). Understanding women's educational and occupational choices. Psychology of Women Quarterly, 18

Maxwell, J. (1996). Qualitative Research Design: An Interactive Approach. Sage Publications. Thousand Oaks, CA.

Seymour, E. and Hewitt, N. (1997). Talking About Leaving: Why Undergraduates Leave the Sciences. Westview Press. Boulder, Colorado.


Home