Why have the proportions of women earning undergraduate computing degrees, and working in the computing workforce, been dropping since the mid-1980s—when women’s participation in nearly all other technical disciplines is on the rise? This question is of great concern to educators, employers, funding agencies, and the U.S. federal government because a sizable, diverse, and creative information technology workforce is critical for continued participation in the high-tech, global economy. Contributors to this volume aimed to shed light on this question, if not to answer it completely, providing insights into possible causes of the current situation and outlining ways to reverse this trend.

It is important to remember that women are scarce only in some aspects of computing. To think of “computing” as a single profession, as such, hides the richness and complexity of the true situation, and this framing may also obscure solutions. For many decades, women have comprised the majority of many low-status, low-paying segments of the computing workforce such as data entry and word processing. Conversely, women have persistently been underrepresented in high-status, high-paying segments such as hardware design and upper level management. There are many professional layers within computing, each with its own distinct story.

This volume identifies several of these layers, and tells the story of each layer’s evolution with respect to gender roles over time. By following the evolution of several segments of computing, in several countries, and identifying the multiple forces that shaped them, we can begin to understand the forces shaping the current situation and how those forces might be realigned.

The primary areas of recent concern are the falling proportions of women in undergraduate computer science programs, and the white-collar professional computing jobs for which those degrees prepare them, such as software developer (programmer) and systems analyst. In undergraduate computer science programs the percentage of women earning bachelor’s degrees fell dramatically from its peak in 1984 at 37% women to less than half this level. Furthermore, the 2007 HERC/HERI freshman interest survey indicates that it is likely to fall further by 2010, possibly leveling out at 10% women [1]. These proportions are similar to the gender proportions in the fields of electrical and mechanical engineering, the most strongly male-dominated of the technical disciplines.

In white-collar computing jobs, there has also been a similar drop that roughly parallels the one in undergraduate education. The proportion of women employed as systems analysts and software developers fell from a peak of 38% in 1987, 3 years after undergraduate enrollments peaked, down to 29% in 2005. The recent losses in female bachelor degree graduates since 2002 are not yet reflected in the workforce, but one can expect that they will be.

These demographic changes are substantial, and they are entirely unlike the patterns observed in other fields of science, technology, engineering, and mathematics (the so-called STEM fields). According to National Science Foundation data, the proportion of women bachelor degree graduates in STEM fields has steadily been rising over the past four decades, from 25% women in 1966 to 51% in 2006. Moreover, women are not disappearing from all segments of computing. In the United States, many of the low-status computing tasks such as data entry and word processing continue to be largely female (see Chapters 3 and 4). While this is not necessarily good news, there is good news in some high-status computing areas. The proportion of women is actually increasing among computer science doctorates and full professors [2]. We emphasize this to combat the common but oversimplified view that women are disappearing from all sectors of computing. They are not.

POSSIBLE EXPLANATIONS

There are numerous hypotheses as to why the proportion of women is shrinking in several important segments of computing. We will explore the strengths and weaknesses of three hypotheses: a lack of female role models; an unappealing, masculine nerd culture; and negative, masculine stereotypes of computing.

Lack of Female Role Models

One could argue that a lack of female role models in computer science is a major contributing cause. Thus, one might expect that increasing the proportion of women faculty in computer science might increase undergraduate women’s interest in signing up for a computer science major. In fact, the proportion of women among computer science faculty has been increasing for the last 20 years, yet the proportion of women undergrads has fallen drastically. Increasing faculty role models has clearly not been sufficient, by itself, to increase the proportion of women undergraduates. There must be additional forces at work.

Masculine Nerd Culture

The male-dominated “nerd” culture is often blamed for chasing women out of computing. However, a male-centric culture alone does not explain why the proportion of undergraduate computer science women is decreasing; there are many other male-dominated “nerd” cultures in fields such as engineering and physics, and yet the proportion of women in these fields continues to increase, not decrease. So perhaps the more appropriate question would be: Is the culture of computer science (CS) more hostile to women than that of other STEM fields? If CS culture is more hostile, then one would expect to see a disproportionate drop-off in the proportion of women in computer science as they advance in education level and career stage. Figure 13.1 shows that the drop-off in computer science is average compared to other STEM fields.

Figure 13.1 shows the difference between the predicted proportion of female faculty and the actual proportion. The “actual” is based on Nelson’s faculty data for the top 100 departments in various STEM fields. The “predicted” is based on the historical pipeline of Ph.D. graduates and is calculated from the National Science Foundation data tracking Ph.D. graduates over the previous 35 years (1966–2001). Fields with the smallest proportions of women, such as the engineering disciplines, appear to be the best at recruiting and retaining women with doctoral degrees into the faculty ranks. Furthermore, fields such as psychology and the biological sciences, which have graduated relatively large proportions of women with doctoral degrees for a long time, do not appear to be particularly successful at recruiting and retaining these women as faculty [3]. This figure raises many additional questions such as: Are engineering disciplines making a greater effort to recruit and retain women faculty? Clearly, additional investigation is needed.

However, the key point for this discussion is that the “shrinkage” of women with increasing rank is no worse in computer science than in the majority of STEM fields. Women are just as likely to advance to faculty ranks in CS as in other STEM fields.

Negative Stereotypes

The “computer science geek” is typically portrayed as an antisocial white male, highly skilled and intelligent, with little attention to personal hygiene. This image is not terribly appealing to either men or women, but it is likely more unappealing to women. The geek image does not necessarily match reality; while such people exist, the average person actually in computer science is not like the “geek.” Thus, it may be external perceptions of computer science culture that deter many women, more so than the actual culture.

The computer geek image has been around for many years. Why should women be more deterred by it now? Indeed, for 20 years, smaller and smaller proportions of undergraduate women have been choosing computer science as a major, as shown in Figure 13.2. Only 1% of all undergraduate women graduated in computer science in 2006. By contrast, at 5.5%, the proportion of undergraduate men graduating in computer science was approximately the same in 2006 as it was in 1986.

What has changed? Early in computing’s history the general public was not particularly aware of what computer programming was, nor what people in that profession were supposed to be like. This was certainly true in the mid-1960s: “The best bit was that no one knew what I did as it was so new” (see Chapter 5). However, by the early 1980s, the increasing popularity and success experienced by computer science also increased media attention and public awareness of computer science stereotypes. Public and media attention on computing certainly increased since the 1980s as computers have crept into almost everyone’s daily activities.

Thus, what has changed is the public awareness of computing stereotypes. We suggest that negative male-centered media images may have turned increasing numbers of women away from computing careers (and also some men). Chapters 8, 9, and 12 illustrate specifically how newspaper and advertising media have created, perpetuated, and modified computing images and stereotypes. Clearly, more research is needed to determine the timing, prevalence, and nature of computing images in the media.

REVERSING CURRENT TRENDS

If the popular image of computer science is a significant factor in the gender gap, then changing or modifying the popular images may be a crucial strategy. While it may be difficult to erase the already established computer geek stereotype, it may be possible to modify it or augment it with other more positive images of computing.

Fortunately, people can hold multiple, possibly conflicting images and stereotypes of a single profession, simultaneously. For example, in the mid-1990s during the Internet craze and the dot.com bubble, several computing stereotypes coexisted simultaneously including the “evil hacker,” the whiz-kid nerd, and the twenty-something entrepreneur-millionaire who was hip yet so respected that he could get away with wearing a T-shirt and jeans to critical business meetings.

The changing prevalence of each of these images may have a strong impact on career choices. Thus, in the mid-1990s, the positive male computer entrepreneur-millionaire image may have counteracted the negative male computer geek image for men, but likely less so for women, since both images were male. This may explain the upsurge of interest in computer science for both men and women during the late-1990s dot.com bubble, although more so for men. That upsurge ended abruptly when the dot.com bubble burst in 2000.

The lessons are that both positive and negative images can coexist, and that positive male images may positively influence both genders, even if one gender responds more strongly than the other. Is it possible for computer science to introduce new, positive images of computing that can counteract the existing negative ones? And if one is to do so, how can they be introduced effectively?

INTRODUCING NEW IMAGES: APPROACHES TO CHANGE

History has shown that some approaches to change are more effective than others. We discuss two broad categories.

Approaches Based on Existing Stereotypes

Corneliussen in Chapter 8 describes three types of discourse about gender and computing: gender blind, which in effect blames women for making the wrong choices in not taking up computing more actively; masculine, which encourages women to think in new ways about computing and to fit into the established culture; and feminine, which attempts to rewrite the meaning of computing as “less technical” and “more social” in a discourse that aims to make computing more appealing to women—if one believes the stereotypes.

Each of these three approaches relies on and perpetuates existing stereotypes that may be problematic. A “gender blind” approach fails to take into account current and very real social and cultural differences between men and women, and as it attempts to motivate through blame, it is unlikely to inspire anyone. Alternatively, both the “masculine” and “feminine” approaches potentially insult women who are already in computing (Chapters 10 and 11), and may alienate both men and women who are potentially interested in computing but do not fit stereotypes (Chapter 8). Finally, the attempt to recast computer science as nontechnical, taken by the “feminine” approach, is a bit like attempting to recast the ocean as “not really so salty.” It is not an effort destined for success. Whether men or women, people interested in computer science will be interested in it for its technical nature, among its many other appeals.

Gender-Independent Approaches

Gender-independent approaches are those that do not make assumptions about the inherent interests and backgrounds of each gender (such as all women like pink and are afraid of technology, or all men like football and love technology), although they may implicitly leverage demographic differences. Carnegie Mellon University (CMU) has employed a successful gender-independent approach that simultaneously increased the number of women and the diversity of ideas and personalities among the men in their programs (see Chapter 2). Modifying the admissions process was the key in CMU’s effort to increase the proportion of women in their undergraduate computer science program. CMU’s admissions focused less on past programming experience and more on leadership, while keeping grade and test standards high [4]. Differences in background were evened out by making background courses available. This approach is gender independent because it neither aims to recast computer science as less technical, nor does it make assumptions about the interests or backgrounds of either men or women.

While it is not entirely clear why this approach worked or what additional factors shaped the results, the removal of prior programming-experience requirements was certainly important. At that time, men were more likely than women to have had prior programming experience (recently this is changing). The removal of background requirements may have removed barriers for women, while also opening the program up to men from more varied backgrounds. The end result was good for both groups. Whether intentional or not, the program encouraged diversity of ideas as well as better gender balance.

The National Academy of Engineering has identified a gender- independent approach to change the image of engineering [5]. This approach could easily be adapted to computer science. (The ACM’s recent initiative, “New Image for Computing,” aims at similar results [6].) The NAE project was initiated in response to concerns about the adequacy of the U.S. technical workforce, and its lack of diversity, and draws on market research techniques. In general, the public has a limited understanding of what engineers do: images may come to mind of men driving trains, of tinkering with machines while wearing dirty coveralls, or using calculators while sitting at drafting tables. Most people don’t realize that “engineers help shape the future” or “engineering is essential to our health, happiness and safety,” to cite two of the messages tested by the project.

Don Giddens of Georgia Tech, chair of the committee that produced the NAE report, states that “we want to emphasize how an engineering career provides an opportunity to change the world rather than over-emphasize the obvious need for strengths in math and science” [7]. While this approach emphasizes the human and social aspects of engineering, which may well be of interest to women at this time in history, it is gender independent in that it does not assume that only women care about these aspects of engineering. Such strategies can increase the diversity of interests found in both women and men in the student body. Moreover, since the human and social aspects are inherent to engineering (if underappreciated by the larger public), this is not simply recasting engineering to appear more “friendly” to women. Georgia Tech emphasizes to prospective students that “engineers change the world.” While many factors likely influence a student’s decisions to come, Georgia Tech graduates a very large proportion of women in engineering [8].

Examples of Image-Changing Strategies

People outside computer science, including young women considering career choices, get their impressions of what people in the profession are like through many sorts of media images in movies, advertising, newspapers, television, and promotional materials. Images may come in the form of written stories, depictions of characters in movies, or photographic images on web sites, advertisements, or other places. Images of many types do have an impact on the implicit biases that people hold. People may or may not be consciously aware of their own biases, yet they impact the way in which people make decisions [9].

However, images can also be used to change these biases. The Implicit Association Test assesses a person’s implicit associations between paired concepts, like “race” and “criminal,” or “race” and “president.” Malcolm Gladwell (a mixed race Jamaican-American) describes how repeated daily exposure to positive images of African Americans improved his IAT scores, indicating that he now held more positive associations with African Americans [10]. More positive and more female images of computing might be used in the same way to change the public view of computer science.

While it may not be possible to directly influence the computing images that commercial advertisers, the press, or Hollywood choose to use, professional computing organizations, government organizations (such as the National Science Foundation), universities, and companies employing computing professionals can choose the computing images they use in their advertising and promotional materials and web pages. Art is another interesting vehicle for presenting images that change people’s viewpoints because of its power to reach people at an emotional and concrete level. Nancy Johnson exhibited portraits of women engineers in 2006 at the Phibbs Center for the Arts in Hudson, Wisconsin. These portraits will eventually be housed near the engineering and science dean’s office at the University of Minnesota, where students, faculty, and staff can see them on a daily basis. The same can be done for computer scientists. The public, especially young people who have not yet made career choices, needs to be exposed to such images, whether contemporary or historic. It is equally important to change the impressions that men have of computer science if one is to change the overall culture or the impressions held by society as a whole.

However, such images additionally have to be placed in additional forums where young people can see them regularly: on YouTube, in coffee shops that house art exhibits, or as a rotating exhibit hung in the hallways of K-12 schools (not just high schools). Through exposure, attitudes might be changed.

The Importance of Local Change

Even when one cannot change the whole of society, one can often change one’s local institution or department. Social change happens through many avenues, including individual and local changes. It is also important to keep in mind that numerical change is not the only important goal; cultural changes that help women (or other underrepresented groups) to feel comfortable and productive in their environment are also important. Significant cultural changes may be possible in a specific department or group, even when achieving gender balance is not.

PATHS FROM THE PAST TO THE FUTURE

Much remains to be done. The studies and analysis of computing’s history contained in this volume provide a better view of the sometimes forgotten but bold roles that women have played throughout computing’s history, for example, as the world’s first programmers and as business entrepreneurs who found opportunity in risks that most men were not willing to take. They have also provided insights into how computing’s current “shrinking women” crisis came about and may provide important guidance in identifying corrective strategies. However, more historical investigation needs to be done in order to truly understand how we arrived at the present. This knowledge is power in shaping a healthy and competitive future for computing [11].

Some of the research that still needs to be done includes further investigation of the hypothesis that rising public awareness of male “computer nerd” and “evil hacker” images factored into the recent 20-year decline in the proportion of women. This will require studies of computing images in news and advertising media, as well as historical interviews with people from the general public to understand computing awareness, attitudes, and perceptions of the general public (e.g., people outside computing).

It is also important to investigate why both women and men entered (or left) computing at various times in history. For example, looking at Figure 13.2, one can see that the recent 20-year shift in the undergraduate gender balance resulted both from a smaller share of women and a larger share of men choosing CS. It is important to understand what affected the choices of both of these groups. While it is hard to get a comprehensive picture of how many women were in computing before the National Science Foundation started to track degrees in 1966, or the Bureau of Labor Statistics started to track computing jobs in 1971, it is still important to understand the dynamics and attitudes of this time.

Historical studies may help change the existing image of computing. Stories and profiles of women in computing gathered through oral history interviews can be presented to the students, researchers, and interested members of the public through a rich variety of existing and new-media forms, ranging from photographic exhibits of successful women to videos on YouTube. Such stories may have the power to counteract the misbegotten idea that computing has always been about men, and may help to attract a more diverse group into computing’s future.

REFERENCES

1. John H. Prior, Sylvia Hurtado, Jessica Sharkness, and William S. Korn, “The American Freshman National Norms for Fall 2007.” Higher Education Research Institute, Graduate School of Education & Information Studies (University of California, Los Angeles, 2007).

2. National Science Foundation, Division of Sci­ence Resources Statistics, Science and Engineer­ing Degrees: 1966–2006. (Arlington, VA: National Science Foundation, 2008); National Center for Women and Information Technology, NCWIT Scorecard 2007 (Boulder: University of Colorado, 2007).

3. Both civil engineering and electrical engineering are more successful in hiring women than is biology, according to 1999–2003 data from the National Academy of Science; see Neil Munro, “Science Faces Title IX Test,” National Journal Magazine (4 July 2009).

4. Lenore Blum and Carol Frieze, “The Evolving Culture of Computing: Similarity Is the Difference” Frontiers: A Journal of Women Studies, Vol. 26, No. 1 (2005): 110–125.

5. National Academy of Engineering, Commit­tee on Public Understanding of Engineering Mes­sages, Changing the Conversation: Messages for Improving Public Understanding of Engineering (Washington, DC: National Academies Press, 2008).

6. For a preliminary report on the ACM’s initiative “New Image for Computing,” see www.acm.org/membership/NIC.pdf (accessed August 2009). This initiative originally focused on differences between racial/ethnic groups, but its preliminary findings are that gender is the more fundamen­tal determinant of positive views of computing. Specifically, “we found relatively small differences in the responses of Hispanic, African American, and White boys, but the disparity between boys and girls is profound” (p. 7).

7. “Changing the Public’s View of Engineering”; available at www.coe.gatech.edu/feature/02_nae.php (accessed August 2009).

8. Karl W. Ritzler, “Tech Urges Women to Try Engineering,” Atlanta Journal-Constitution (2 May 2008).

9. A. G. Greenwald, D. E. McGhee, and J. L. K. Schwartz, “Measuring Individual Differences in Implicit Cognition: The Implicit Association Test,” Journal of Personality and Social Psychology, Vol. 74, No. 6 (1998): 1464–1480.

10. Malcolm Gladwell, Blink: The Power of Think­ing Without Thinking (New York: Little Brown and Company, 2005).

11. D. A. Lenat and E. A. Feigenbaum, “On the Thresholds of Knowledge,” Proceedings of the International Joint Conference on Artificial Intelligence, Vol. 2 (1987): 1173–1182.