President's Message: Cherchez la Femme
Over that past year and a half, it has been easy to write about all of the things we can celebrate about signal processing: an inspiring 50-year history, thriving new ventures in multimedia and communications, exciting prospects for electronic publications, yet another successful ICASSP, and a bright (and maybe even well-funded) future at the confluence of computing, communications, and signal processing. Are there any clouds on the horizon? I think that, in fact, there is an area where we are falling short: as a field, we lack diversity along several dimensions, with one of the most noticeable, from my point of view, being the number of women in signal processing.
I don't know if most signal processors notice the lack of women. I do remember, as a graduate student, attending my first conference with my Ph.D. thesis advisor. Part way through the conference he turned to me and said ``You know, there aren't many women here. I never noticed that before.'' I would assert that noticing is, in fact, the first important step -- a step that then leads to three key questions:
I'll start with the good news: the underrepresentation of women isn't really a signal processing problem. It's a problem we've inherited from our larger context, namely engineering as a whole and electrical engineering in particular. As reported by the American Association of Engineering Societies (AAES) Engineering Workforce Commission, the 1998 numbers for the United States show 18.6% of the 75,058 Bachelor's degrees going to women, 20.3% of the 36,337 Master's degrees, and 12.3% of the 7377 Ph.D degrees. Across 20 engineering fields, EE ranks in the bottom four, with 12.3% of Bachelor's degrees awarded to women. (The others at the bottom are Petroleum Engineering, also with 12.3%, Mechanical Engineering, with 12.2%; and Marine/Naval Architecture, with 11.9%. With EE and ME accounting for over 40% of the Engineering Bachelor's degrees, the numbers don't look good. And in case you're wondering, Environmental Engineering is at the top of the list, with 37.8% of Bachelor's degrees going to women.) In 1995, women represented roughly 10% of the engineering work force in the U.S. In some countries -- e.g., China, Italy, Turkey, Hungary -- women are somewhat better represented, but world-wide, women account for a small percent of the engineers.
I don't think anyone collects official numbers for signal processing. Women were pretty invisible in the governance of the Society for the first 25 years of our 50-year history. At first glance, the picture for women in Society leadership positions over the most recent 25 years looks pretty respectable: two Presidents, one Vice President-Conferences, two past Vice Presidents-Awards & Membership, one past VP-Finance, one Secretary-Treasurer, two ICASSP General Chairs, a past Editor-in-Chief of the Transactions on Signal Processing and a past Editor-in-Chief of the SP Magazine. Why, in 1998, three of the eight Society officers were women! The record looks somewhat less impressive, however, when you realize that this long list of activities represents a total of only five women, and that one amazing woman, Delores Etter, accounts for five of those 11 positions. Counting women elected to the Society's AdCom/Board of Governors over the past 25 years adds only five new names, bringing the grand total to ten. Active, yes; plentiful, no. From top-of-the-head enumerations of known faculty and students, informal observation at ICASSP, and a quick run-through of past and present colleagues active in the SP Society, though, our demographics look neither better nor worse than those for EE as a whole. We can therefore console ourselves that signal processing probably isn't the problem. But what is?
There is a wealth of literature from fields such as education and psychology that explores subtleties of the relationships between gender and math, science, and engineering. The following is a (non-uniform) sampling.
Girls and math: Past success in math and encouragement of parents and teachers can lead to more positive attitudes about math, more persistence in pursing math education, and great achievement in math [Sherman, 1983; Eccles and Jacobs, 1986; Ethington and Wolfe, 1986] (hardly a surprise). Different studies can come to opposite conclusions -- e.g., about whether or not there are gender differences in math anxiety at various ages. Other than spatial abilities (probably greater for boys than for girls), there is no solid documentation of greater math abilitites for either girls or boys [Freedman, 1989; Sadker, Sadker, and Klein, 1991]. A study that concludes that boys outperform girls in problem solving also concludes that girls outperform boys in understanding mathematical concepts [Hyde, Fennema, and Lamon, 1990]. A compilation of over 100 studies that included 3,000,000 test scores concludes that the small differences that have, in some studies, been recorded have decreased over time, suggesting environmental rather than genetic factors [Hyde, Fennema, and Lamon, 1990].
Reasons aside, however, girls take fewer math classes in high school [Fennema, 1985] and perceive math as less useful than boys [Brush,1985]. I recently asked two different groups -- 100 10- and 11-year old Girl Scouts and 40 high school girls -- to rate the importance of various subjects for a career in engineering. Both groups thought math was important and language (in their case, English) unimportant. The Girl Scouts rated math and language as having equal (un)importance: only 10% of the Scouts thought either was important for engineers.
Education: There is a large body of evidence that, from preschool through grad school, teachers -- both male and female -- interact differently with male and female students [Sadker and Sadker, 1994]. The difference that has the most obvious bearing on the number of women in engineering is the failure to encourage girls to take advanced math and science classes in high school [Cosgrove, Blaisdell, and Anderson, 1994; Brown, 1995]. One of my (female) colleagues tells of hearing her daughter's high school math teacher urge other parents to encourage their sons join the math team, but never saying a word about the math team to her daughter, even though her daughter had by far the highest math scores in the class.
Self-efficacy: Self-efficacy refers to a person's belief that s/he can perform a given task [Bandura, 1997]. Studies in self-efficacy draw some fairly intuitive conclusions -- e.g., that success in an endeavor increases self-efficacy ratings [Maddux and Stanley, 1986] -- but also report some less predictable findings. In several studies, failure in an endeavor tended to cause a greater decrease in self-efficacy ratings of women than of men [Hackett and Campbell, 1987]. (An example of this is related by Jane Daniels, director of Purdue's Women in Engineering Programs. After the first round of freshmen tests, the freshmen counselors report that they hear very different stories from the female and male students who did poorly. The women will question their own ability and their future in engineering; the men will blame the "stupid professor" for giving such a bad test.) Also telling is a study that asked undergraduates to assess their ability to complete courses of studies for 20 occupations [Betz and Hackett, 1981]. The male students rated themselves as capable of completing the studies for an equal number -- 6.9 -- of the traditionally male and traditionally female jobs; the female students rated themselves as capable of completing the studies for 8 of the traditionally female jobs and 6 of the traditionally male jobs. But whereas men rated the occupation of physician as most difficult, women gave that distinction to engineering.
The direct approach: A few studies have attempted to tackle the question head-on. A recent survey of (male and female) leaders conducted in conjunction with the May 1999 National Academy of Engineering's Summit on Women in Engineering rated principal factors at key stages of education and career -- pre-college, college, and professional. The one factor cited at all three levels was lack of role models. Other factors influencing pre-college (roughly ages 5-18) students included lack of awareness of what engineers do, failure to portray engineering as a viable career for women, failure to encourage girls to take advanced math and science courses, and negative images of engineers (the computer engineering in Jurassic Park was not the good guy). At the college level, lack of role models is joined by lack of mentors, isolation, and lack of professional/academic networks as the factors cited most often, followed by negative reactions to introductory math or science courses, a hostile academic environment, and lack of self confidence. At the professional level, many of the themes are repeated: lack of role models and mentors, isolation, lack of professional networks, a hostile work environment. However, there are also new themes: conflict between the demands of workplace and family, differential treatment, lack of recognition/rewards/incentives.
For a capsule summary, a 1995 University of California-Davis study involving women engineering students, faculty, and professionals [Henes et al., 1995] provides one of the clearest pictures. The UC-Davis researchers found five principal reasons why women leave or become discouraged with engineering. The factors are those identified in the study; the elaborations are mine.
So that's the picture. For why it matters and what can be done, tune in next time. There's no silver bullet, but there are some points of light. Probably the best news is that changes in pedagogy and culture that make engineering -- and signal processing -- more welcoming for women are likely to improve the environment for everyone.
Leah H. Jamieson