First, we’ll review the common explanations given for this situation and the formal research that investigates them.
Much research has been done on innate ability differences, preferences, and cultural biases as reasons for the underrepresentation of women in science, technology, engineering, and mathematics (STEM) fields. Ceci, Williams, and Barnett developed a framework to understand how these all interact [Ceci et al. 2009]. Next, we address the research on each factor and then work it through Ceci et al.’s more integrative framework. The picture that emerges (see Figure 13-1) gives the reader a feel for the complexity of the interactions between the contributing factors. Although there are certainly biologically rooted gender differences at work, the research suggests that there also may be some detrimental gender biases involved, which raises further questions.
Innate ability differences between males and females (as well as environmentally mediated differences traceable to experiences during childhood) have been explored as one possible reason for the declining number of women in computer-related fields. Substantial evidence supports the argument that women are not as capable at highly math-intensive pursuits as are men. This sex asymmetry is found at the very upper end of the ability distribution. For example, the top 1% of scores on the mathematics SAT shows a 2-to-1 ratio of males to females, and the top .01% shows a ratio of 4-to-1 [Hyde and Lynn 2008]; [Lubinski et al. 2001]. Males also earn most of the very low scores, meaning that males’ performance is simply more variable overall.
Ceci, Williams, and Barnett [Ceci et al. 2009] divide the evidence on cognitive sex differences into mean differences (at the midpoint of the distribution) and right-tail differences in proportions in the top 10%, 5%, and 1%, the latter being a better representation of those in the science, technology, engineering, and math (STEM) professions. Based on a national probability sampling of adolescents between 1960 and 1992, Hedges and Nowell found that the distribution of test scores for male and female test-takers differed substantially at the top and bottom 1%, 5%, and 10% [Hedges and Nowell 1995]. Males excelled in science, mathematics, spatial reasoning, social studies, and mechanical skills. Females excelled in verbal abilities, associative memory performance, and perceptual speed. These findings raise the possibility that biology accounts for some of the observed gender patterns of participation in related fields of STEM, CS, and IT.
Research on relative brain size, brain organization, and hormonal differences is also relevant. Ceci and Williams review the recent biological work on cognitive sex differences, investigating brain size, brain organization, and hormonal differences [Ceci and Williams 2010]. Discussing Deary et al.’s finding of a modest correlation (.33–.37) between intelligence and brain volume [Deary et al. 2007], in which men on average have slightly bigger brains, Ceci and Williams note that “in most of the research on biological correlates of sex differences, the focus is on means, whereas the focus on sex differences in the STEM fields is on the extreme right tail (the top 1% or even the top .1% or the top 0.01%).” In other words, many studies of average brain differences are not pertinent to our question, because strong evidence of mathematical and spatial ability differences between men and women appear only at the very top (or bottom) of the range of ability scores.
Other research cited in Ceci and Williams’ review suggests that males and females use different parts of their brains to complete the same tasks [Haier et al. 2005]. Ceci and Williams conclude that “with additional independent replications and representative sampling, it can be concluded that men and women achieve the same general cognitive capability using somewhat different brain architectures.”
Additionally, Ceci and Williams cite research that investigates the role of pre- and postnatal hormones in understanding cognitive sex differences. In one study, male rats were superior at figuring their way around a maze, compared with female rats. Once the male rats were castrated, their superiority disappeared. Ceci and Williams also review research in which biological females, given estrogen-suppressing drugs coupled with large doses of male hormones during sex-change operations, developed enhanced spatial abilities. The large body of research in this area suggests that hormonal factors might affect professional choices of women. However, it is unclear how much. Ceci and Williams conclude that the evidence is “not strong and consistent enough to justify claiming that hormones are the primary cause of sex differences in STEM careers.”
Before we leave the subject of hormonal differences, however, we should consider the possibility that they underlie some behavioral differences that predispose women not to be as attracted as men to working in computer science.
Statistics show that women are committed to the professional work force. They hold 57% of all professional occupations in the U.S. in 2008 [Ashcraft and Blithe 2009]; [National Center for Education Statistics 2008], and they are also successful in math (as measured by grades), a closely related academic discipline. Thus, it seems important to go beyond the explanation of ability deficits and to ask about women’s choices. The statistics call for a gender-sensitive analysis of the factors influencing women’s decisions to participate in the field of Computer Science—or not—and we also need to address the possibility that women find themselves disenfranchised by the male culture of CS. If, in fact, significant reasons for a gender imbalance lie here, then here, too, may exist an opportunity to reverse a portion of this trend.
Accordingly, some researchers have addressed preferences and cultural forces. Some claim that culturally inscribed career and lifestyle choices are the major reason for the small number of women in computer science, and others claim more strongly that discouraging cultural forces are the most instrumental causes. Next, we review evidence for each of these positions.
With respect to career choice, gender shifts within professions have occurred throughout history, notably within teaching, secretarial work, and medicine [Ceci and Williams 2010]. These shifts are easily explained by changes over time in these careers’ prestige levels and financial remuneration, rather than by hormones or genes. Repeatedly, men have taken over whatever kind of work is considered more economically valuable, suggesting that gender workforce patterns are driven more by cultural and political forces rather than simple biological differences. In a recent longitudinal study of women’s choices to work in health-related careers, we can find an interesting parallel case in which cultural values drive career choices. Jacqueline Eccles and colleagues at the University of Michigan found that even when mathematical ability was taken into consideration, young women were more attracted to health-related careers because they placed a higher value on a people/society-oriented job than did their male peers [Eccles et al. 1999].
Margolis, Fisher, and Miller [Margolis et al. 2000] provide further evidence of a “female” inclination—or values choice—to serve people and society in their 2000 study involving 51 male and 46 female computer science majors at Carnegie Mellon University (comprising a total of 210 interviews). A representative quote from a female computer science interviewee resonates with Eccles’s research:
The idea is that you can save lives, and that’s not detaching yourself from society. That’s actually being a part of it. That’s actually helping. Because I have this thing in me that wants to help. I felt the only problem I had in computer science was that I would be detaching myself from society a lot, that I wouldn’t be helping; that there would be people in third-world countries that I couldn’t do anything about...I would like to find a way that I could help—that’s where I would like to go with computer science.
Margolis, Fisher, and Miller found that women’s quest for people-oriented purposes for computers was in concordance with other research in the field of computer science [Honey 1994]; [Martin 1992]; [Schofield 1995]. They report that 44% of the female students in their study (as compared to 9% of the male students) emphasized the importance of integrating computing with people through projects with a more human appeal. Overall, women preferred computing for medical purposes (e.g., pacemakers, renal dialysis machines, and figuring out diseases), communication, and solving community problems over computing for the sake of computing, developing better computers, or programming for games.
Tagging some similar values issues, Ferriman, Lubinski, and Benbow point to gender differences in lifestyle preferences and orientation toward life as the main reason for women’s underrepresentation in high-intensity STEM careers [Ferriman et al. 2009]. Their research is unique in that they were able to hold ability constant and narrow the population down to only those who excel in STEM careers. By following mathematically precocious youth over 20 years, they found that “following the completion of their terminal graduate degrees, men seem to be more career-focused and agentic, whereas women appear to be more holistic and communal in their orientation toward life and more attendant to family, friends, and the social well-being of themselves and others more generally.” By this argument, then, there are few women in CS simply because women are more interested in and prefer other disciplines and areas.
Some researchers reject the notion that any inherently female quality (whether ability or interest) causes women’s underrepresentation in CS and IT careers. They argue instead that the culture of CS and IT discourages women. In “The Anatomy of Interest: Women in Undergraduate Computer Science,” Margolis, Fisher, and Miller focus on how women students who enter CS with high enthusiasm and interest in computing quickly lose their ability and interest in the subject [Margolis et al. 2000]. They looked at factors beyond intellectual preference that influenced interest in an abstract body of knowledge. For example, they explored how gender-biased norms eroded confidence, and also how a masculinized standard for success shaded women’s interest and ability in computing. The authors suggest that there may be some “pernicious ways in which male behavior and interest become the standards for ‘the right fit’ and success,” and this, in turn, contributes to women’s waning enthusiasm in the subject. In other words, as their interviews showed, women who refused to conform to the image of the myopically focused “computer geek” who “hacks for hacking’s sake” might feel out of place.
For those who perceive the culture of computing as one in which the “boy wonder” icon is up all night programming feverishly in isolation, Margolis, Fisher, and Miller offer this insight from a female computer science teacher:
My point is that staying up all night doing something is a sign of single-mindedness and possibly immaturity as well as love for the subject. The girls may show their love for computers and computer science very differently. If you are looking for this type of obsessive behavior, then you are looking for a typically young, male behavior. While some girls will exhibit it, most won’t. But it doesn’t mean that they don’t love computer science!
Shortcomings of the Margolis, Fisher, and Miller case study include the fact that it examines just one small subset of the general population of students pursuing computer science, and thus, we should be wary of extrapolating these personal accounts to the broader population. We should not make broad assumptions based on this small sample. Furthermore, even though their interview questions were designed to elicit students’ own experiences rather than their abstract thoughts, the authors admit that this interviewing technique was not conducive to assigning relative weight to different detachment factors, as “factors frequently shifted and appeared enmeshed with one another” [Margolis et al. 2000].
At the same time, these findings resonate with other studies of computer culture, such as one by the Educational Foundation of the American Association of University Women (AAUW), which combines input from its 14 commissioners (researchers, educators, journalists, and entrepreneurs) in cyberculture and education. Their report covers the Foundation’s online survey of 900 teachers, qualitative focus research on more than 70 girls, and reviews of existing research, in order to provide insight into perspectives on computer culture, teacher perspectives and classroom dynamics, educational software and games, computer science classrooms, and home community and work [AAUW 2000]. Like Margolis, Fisher, and Miller, the AAUW found cultural deterrents to female participation in computer science. They found that girls are concerned about the passivity of their interactions with the computer as a “tool.” Additionally, they found that girls rejected the violence, redundancy, and tedium of computer games and expressed dislike for narrowly and technically focused programming classes. Furthermore, the AAUW contends that these concerns are dismissed as symptoms of anxiety or incompetence that will diminish once girls “catch up” with the technology.
Finally, in a comprehensive compilation of research in IT, CS, and CE, McGrath Cohoon and Aspray integrated research from over 34 key researchers in the field [McGrath Cohoon and Aspray 2006]. Their potential explanations for the underrepresentation of women include experience, barriers to entry, role models, mentoring, student-faculty interaction, peer support, curricula, and pedagogy, as well as student characteristics such as academic fitness, values, confidence, and response to competition, plus the culture of computing.
In light of these culturally based concerns, we might ask what, exactly, high-ability women who opt out of disciplines such as CS do choose to do with their intellectual lives? Ceci, Williams, and Barnett remind us that women with high math competence are disproportionately more likely than men to also have high verbal competence, allowing them greater choice of professions [Ceci et al. 2009]. Hence, issues of culture and choice likely dovetail, directing capable women out of the computer field, thus revealing that more than biology, and factors other than raw ability, are at play. Figure 13-1 depicts the interplay of all these factors, both biological and cultural.
With so many confounding factors, it is no surprise that we have no clear solution to the barriers that some women may face in CS and related fields. On the other hand, we do have an emerging picture of multiple and interacting forces potentially acting against women’s full participation, which raises implications to which we now turn.