In general, while female representation in the Faculty of Science has increased, the representation of females in Computing Science has not. Only 24 of the 1349 Faculty of Science females were enrolled in Computing Science programs in 1997. The percentage of females in Math remains consistently higher than in Computing Science, although in recent years, the absolute numbers are also dwindling.

Figure 3 plots the corresponding data for students enrolled in the Faculty of Engineering and in the Computer Engineering program. There has been a marked increase in the number of females entering Engineering as a discipline at the University of Alberta during this time period. The Computer Engineering program, capped by an enrollment quota, accounted for about 8% of the engineering students during this time frame. While Engineering has enjoyed an influx of females into the discipline as a whole, this does not seem to be the case for Computer Engineering. The percentage of Computer Engineering females averages about 8% of the total students enrolled in this program, and again, the absolute numbers are quite low. For the years 1990, 1993, and 1997, the number of females enrolled in the Computer Engineering program were 13, 10, and 17, respectively. The total number of Computer Engineering students for these years were 145, 144, and 179.

4. Performance in Introductory Computing Science Courses

The data presented above indicate clearly that female enrollment in computer science degree programs at the University of Alberta is declining. We can then consider two possibly related questions: (a) Are increasingly fewer females even trying to enter the program? and (b) Is there a significant difference in performance between females and males in the introductory courses required for admission into the program? The enrollment and average grade data for females and males in three courses–CMPUT 114, CMPUT 115, and ENCMP 100– were collected for 1992-1997, inclusive. CMPUT 114 is the first course that any student contemplating a computing science degree program must take and it has been defined as an introduction to programming and problem-solving. It is followed by CMPUT 115, a course that focuses on abstract data types and programming problem-solving with more advanced data structures. These two courses are open to all university students, but the majority of students are Faculty of Science students who either anticipate applying for entrance into the computing science program or who take the course as an elective or requirement for other science programs. There is a small percentage of students each year from other faculties. ENCMP 100 is a course analogous to CMPUT 114 and is a required course for all first-year Engineering students, regardless of what engineering sub-discipline they subsequently pursue at the start of their second year.

Enrollment and Completion for Prerequisite Courses. We can address the first question–Are fewer females trying to enter the computing science program?– using the number of females enrolling CMPUT 114 as a proxy for "interest" in computing science and an initial intention to pursue this degree. In 1992, about 33% of the students enrolling in CMPUT 114 were female (90/276). In 1997, this figure was 20% (77/370). Keeping in mind that the number of females in the Faculty of Science was increasing during this period, the apparent trend in interest in computing science is not a promising one.

We can then consider the number of females and males who complete the second required first-year course, CMPUT 115. Because completing is not the same as passing, the completion figure is an optimistic upper bound on the number of students who might be eligible to enter the computer science program. In 1993, 21% of the students completing CMPUT 115 were female (27/98). In 1998, this figure was 15% (42/244).

To complete this picture, we can examine withdrawal rates from these courses as a function of gender. For the years 1992-1997, the average withdrawal rate for females in CMPUT 114 was 21.2%; for males, it was 16.5%. For CMPUT 115, an average of 27.4% of the females withdrew; For males, the rate was 21.5%. In sum, proportions of females enrolling in and completing the prerequisite computing courses for the computing science degree program are declining over time.

Performance in Prerequisite Courses. At this point, we can turn to the question of whether there is a difference in grades achieved by females and males in these introductory courses. The University of Alberta uses a 9-point grading system. On this scale, a grade of 4 is the minimum passing grade in a course; grades of 6-7 are considered "very good" and grades of 8-9 are considered "excellent." It is not be possible to determine, using the aggregate data available for this study, whether any difference in the average grades achieved by females and males in these courses is statistically significant. However, as someone who has taught both these courses and other large introductory courses, I think it fair to say that a 0.2 difference does not signal any practical difference in performance; 0.4 difference is probably signaling a real difference, and any difference greater than 0.4 certainly is. As context for interpreting what a particular grade means in terms of overall performance in a course, a 1997 section of CMPUT 114 assigned a final grade of 6 to students whose overall term performance fell between 66% and 69.6% out of 100%. In this same section, students receiving a grade of 5 had a final term performance grade between 58.8% and 65.65% out of 100%.

Figure 4 shows that females consistently achieve a lower grade in CMPUT 114, that this difference is never smaller than 0.3 of a grade-point. For several years, the gap was considerably larger, reaching as much as a 0.9 grade-point difference in 1994. Figure 5 plots the analogous data for CMPUT 115. In 1993, 1996, and 1997, it is fair to say there is no real difference in the average grades for the two groups. In the remaining years, there was a sizable gap ranging from 0.4 to 0.9 of a grade point , with females achieving the lower grade on average.

We must take care in any conclusions drawn from these averages, because the sample sizes on which these means are based are radically different for females and males, and we do not have any variance information. For example, the 1997 mean grade of 5.7 for females in CMPUT 114 is based on 64 students; the mean grade of 6.2 for males is based on 246 students.

Finally, Figure 6 presents the average grade in ENCMP 100 for females and males. Interestingly, there is no practical difference in GPA's achieved by these two groups for this course, which is a somewhat more rigorous version of CMPUT 114 that includes emphasis on numerical methods and precision as it relates to engineering problems. Engineering students must successfully complete ENCMP 100 to proceed in specialized programs within engineering. Not surprisingly, the average withdrawal rates from ENCMP 100 are considerable lower than the rates for CMPUT 114 for the years 1992-1997 and have been lower for females than for males (4.8% for females, 5.4% for males ).

To summarize, the number of females choosing and completing introductory computer science courses offered in the Faculty of Science is declining in both absolute and relative terms. A higher rate of females than males withdraw from these courses at the University of Alberta. In the first required computing course, females consistently achieve lower grades than males and the difference is arguably a real difference, in terms of performance on assigned work. In some years, it has been over half a grade point. These remarks do not hold for the performance of females in the Engineering faculty who take an analogous introductory computing science course offered jointly by Engineering and Computing Science. Now, the students taking ENCMP 100 are a more homogeneous group than are the students taking CMPUT 114. In the ENCMP 100 case, all students are Faculty of Engineering students and have comparable high school backgrounds in order to be admitted to the Engineering faculty. CMPUT 114, on the other hand, draws students from other faculties as well as science students for whom computer science is not an intended major. In short, the two populations have met different admission criteria, have different preparatory backgrounds, and the two introductory courses likely play different roles in the academic programs and priorities for the students taking one or the other.

That said, the grade point average difference between females and males in CMPUT 114 deserves closer examination. If it were only the variability in the populations taking CMPUT 114 v. ENCMP 100, then we would expect that both males and females in CMPUT 114 would perform less well than their counterparts in ENCMP 100. We can only speculate on some possible causes. One might be related to the relatively small numbers of females in CMPUT 114 relative to the males. Success in high workload courses–particularly coding and software courses– can have a social component, i.e., a network of support in which students turn to each other for insight and help on assignments and on studying. It may be that the relatively few females in these large courses are not easily absorbed into networks-of-support that emerge in the labs and after hours. If this is a contributing cause, then steps to facilitate study groups for the "outsiders" –which may include females, students from outside the Faculty of Science, older students of either gender–might address this problem.

5. Enrollment Trends in Graduate Programs

The Department of Computing Science offers both Masters and Ph.D. degrees in computing science. The faculty of Engineering only recently offered graduate degrees in Computer Engineering; prior to 1996, graduate degrees were offered in Electrical Engineering. In 1996, these degree programs became graduate programs in "Electrical Engineering/Computer Engineering."

5.1 Method and Results

The "Summary of Statistics" document prepared by the University of Alberta's Registrar's office also includes demographic data students enrolled in graduate studies at the university. Because part-time status is quite common for students during their graduate career, both part and full-time students were included in the data presented below.

Table 1 presents the numbers of females and males pursuing graduate degrees in computer science and electrical engineering, for the years 1990-1997. The representation of females in the computing science graduate program at the University of Alberta has remained fairly constant during the time-frame examined, and is higher than the number of females in the undergraduate program.

The following features of these data should be underscored. Female students are underrepresented in computing science at the University of Alberta, relative to their representation in the Faculty of Science as a whole and relative to their representation in related disciplines, such as Mathematics. It is unclear why the numbers of females pursuing computing science degrees has been declining, while females are enrolling in science programs at an increasing rate.

Without considering similar data from other universities, we cannot tell whether or not the trends identified at the University of Alberta are unique. And it is impossible to identify any causal factors that might be responsible for the declining numbers of females pursuing computing science degrees at the University of Alberta. Three oft-cited factors for poor representation of females in information-science fields are (a) relatively few female role models, (b) a perception that the computer-related careers are for "nerds," where that label has negative personal and lifestyle connotations, and (c) the lack of computer games that appeal to a broad range of young girls that would serve to increase their computer-related activities and confidence, as young boys are hooked. A consideration of the latter two factors is outside the scope of this report. However, one would expect that, if anything, the opportunities for young females to obtain computer skills and confidence have only increased over the past decade at the primary and secondary school level. And yet we see a decline of females pursuing computer science degrees at the University of Alberta.

During 1990-1997, the number of females faculty members in the Department of Computing Science increased from three to five, to account for 15% of the tenured or tenure-track faculty. However, there were no female instructors teaching the two required first-year computing science courses in the years 1990, 1995, and 1997. For each of the years between 1991-1994 inclusive, one of the three CMPUT 114 sections was taught by a female instructor. Only once during this period did a female instructor each a section of CMPUT 115. The female faculty members do teach core, required courses within the program. However, students do not encounter these instructors until after they have been admitted to the program. Whether or not this has a significant impact on the recruitment of female students to the undergraduate program is difficult to determine.

Special thanks go to Kam Kong, Programmer Analyst in the Office of the Registrar at the University of Alberta, for extracting the grade data for males and females enrolled in the introductory computing courses.